Mumuki_Entrega_Video_2.ipynb

{
  "nbformat": 4,
  "nbformat_minor": 5,
  "metadata": {
    "colab": {
      "name": "Mumuki_Entrega_Video_2.ipynb",
      "provenance": [],
      "collapsed_sections": [
        "ScIzoHTB-uOO",
        "DN_jjsoHb37W",
        "vzKtcQ9ZmDFS",
        "CEpmdp5Rma_V",
        "5CJDLNT5ma_2",
        "nVRXPHjBma_8",
        "jEfV3Mq2mbAB",
        "54jOL7Z1ImKs",
        "PdOC28Ni9vjW",
        "7BN3L7TXlzq1",
        "HZHvFlsfq-vt",
        "BkK68BiRoPz_",
        "z8sIwflVwstO",
        "G0TdNxZ3Z716",
        "xpo4ts_Fffu4",
        "7E-GsFkWnoP2",
        "p6WpNzsP05b4",
        "SfdGqGadpbnp",
        "EFdE7XhgoDX-",
        "OWqIvpWk3Dn-",
        "dXYt9ae4J4YZ",
        "EhsYtHNXIS9_",
        "4oCN_hS7LEOD",
        "5SZmThBxGT53"
      ],
      "toc_visible": true,
      "include_colab_link": true
    },
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "codemirror_mode": {
        "name": "ipython",
        "version": 3
      },
      "file_extension": ".py",
      "mimetype": "text/x-python",
      "name": "python",
      "nbconvert_exporter": "python",
      "pygments_lexer": "ipython3",
      "version": "3.8.8"
    }
  },
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "view-in-github",
        "colab_type": "text"
      },
      "source": [
        "<a href=\"https://colab.research.google.com/github/NadiaFinzi/diplodatos-mumuki-equipo1/blob/main/Mumuki_Entrega_Video_2.ipynb\" target=\"_parent\"><img src=\"https://colab.research.google.com/assets/colab-badge.svg\" alt=\"Open In Colab\"/></a>"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ScIzoHTB-uOO"
      },
      "source": [
        "**Diplomatura en Ciencia de Datos, Aprendizaje Automático y sus Aplicaciones**\n",
        "\n",
        "**Edición 2021**\n",
        "\n",
        "---\n",
        "## Mentoría MUMUKI\n",
        "## **\"Predicción de deserción en una plataforma educativa para aprender a programar\".**\n",
        "\n",
        "### Mentor\n",
        "\n",
        "- Bulgarelli, Franco Leonardo\n",
        "\n",
        "### Integrantes\n",
        "\n",
        "- Finzi, Nadia\n",
        "- Gherra C., Estefanía Inés\n",
        "- Primo, Eugenia\n",
        "\n",
        "### Videos\n",
        "\n",
        "*   [Primera entrega](https://www.youtube.com/watch?v=cM2WSmDDu9Y)\n",
        "*   [Segunda entrega](https://www.youtube.com/watch?v=u7QfSTs7Tis)\n"
      ],
      "id": "ScIzoHTB-uOO"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "DN_jjsoHb37W"
      },
      "source": [
        "# Bibliotecas, funciones e importación de datos"
      ],
      "id": "DN_jjsoHb37W"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "31a27d4f-b545-4b2f-a5d3-9e4f19c1ba34"
      },
      "source": [
        "# Librerías y configuraciones generales\n",
        "\n",
        "# Esenciales\n",
        "import pandas as pd\n",
        "import numpy as np\n",
        "from datetime import datetime, date, time, timedelta\n",
        "import calendar\n",
        "\n",
        "# Visualizacion\n",
        "import matplotlib.pyplot as plt\n",
        "plt.style.use('ggplot')\n",
        "plt.rcParams['axes.facecolor']='w'\n",
        "plt.rcParams['grid.color'] = 'grey'\n",
        "plt.rcParams['grid.alpha'] = 0.2 \n",
        "plt.rcParams['grid.linestyle'] = \"-.\"\n",
        "\n",
        "%matplotlib inline\n",
        "import seaborn as sns\n",
        "\n",
        "# Varios\n",
        "import warnings\n",
        "warnings.filterwarnings('ignore')\n",
        "\n",
        "# Entrenamiento\n",
        "from sklearn.model_selection import train_test_split\n",
        "from sklearn.linear_model import LogisticRegression\n",
        "from imblearn.over_sampling import SMOTE\n",
        "\n",
        "# Métricas\n",
        "from sklearn.metrics import classification_report, accuracy_score, precision_score, recall_score, f1_score, confusion_matrix, plot_confusion_matrix\n",
        "from sklearn.metrics import f1_score"
      ],
      "id": "31a27d4f-b545-4b2f-a5d3-9e4f19c1ba34",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "yjViw0bdkAqz"
      },
      "source": [
        "# Definición de funciones y procedimientos propios\n",
        "\n",
        "def graficar_boxplot(dataframe, columna, aclaracion_titulo, label_x, pie_grafico):\n",
        "  fig = plt.figure(figsize=(8, 6))\n",
        "  g = sns.boxenplot(data=dataframe, x= columna, color='salmon')\n",
        "  plt.title('Boxplot de la variable ' + columna + aclaracion_titulo, fontsize=14, weight='bold')\n",
        "  plt.ylabel(\"\")\n",
        "  plt.xlabel(\"\\n\" + label_x, fontsize=10)\n",
        "  plt.annotate( pie_grafico, xy=(0.3, -0.25), xycoords='axes fraction', ha='right', va='center', fontsize=12)\n",
        "\n",
        "def submissions_correctas(df):\n",
        "  return df[(df[\"submission_status\"] == \"passed\") | (df[\"submission_status\"] == \"passed_with_warnings\")]\n",
        "\n",
        "def submissions_incorrectas(df):\n",
        "  return df[(df[\"submission_status\"] == \"failed\") | (df[\"submission_status\"] == \"errored\")]\n",
        "\n",
        "def ordenar_y_agrupar(df):\n",
        "  return df.sort_values([\"student_uid\", \"submission_created_at\"]).reset_index(drop=True).groupby([\"student_uid\", \"exercise_new_id\"])\n",
        "\n",
        "def plot_scatter_grid(x, y, exercises):\n",
        "  row = 3\n",
        "  col = 2\n",
        "  i = 1 # inicia el conteo\n",
        "\n",
        "  plt.figure(figsize = (20,15))\n",
        "  for exercise in exercises:\n",
        "    selector = estudiante_por_ejercicio[estudiante_por_ejercicio[\"exercise_new_id\"] == exercise]\n",
        "    plt.subplot(row, col, i)\n",
        "    plt.title(\"Ejercicio \" + str(exercise))\n",
        "    sns.scatterplot(x = x, y = y, hue = 'cond_cant_ejer_resueltos', data = selector, facecolor=\"dodgerblue\", edgecolor=\"k\", palette=[\"blue\", \"red\"])\n",
        "    i += 1\n",
        "\n",
        "  plt.subplots_adjust(wspace = 0.3, hspace = 0.5) \n",
        "  plt.show()\n",
        "\n",
        "# Funciones para armar modelos de regresión logística\n",
        "\n",
        "def data_por_student(ids_ejercicios, imputar_con_constante, imputacion_nan):\n",
        "  # Base con desagregación de ejercicios, por variables: intentos_incorrectos y tiempo_total.\n",
        "  data = estudiante_por_ejercicio[estudiante_por_ejercicio.exercise_new_id.isin(ids_ejercicios)][[\"student_uid\", \"cond_cant_ejer_resueltos\", \"exercise_new_id\", \"intentos_incorrectos\", \"tiempo_total\", \"ultimo_estado\"]]\n",
        "  \n",
        "  data[\"ultimo_estado\"] = 0  # 0 = resolvió el ejercicio, sin importar en qué estado\n",
        "  \n",
        "  # Armamos la tabla pivot\n",
        "  data[\"exercise_new_id\"] = data[\"exercise_new_id\"].astype(str)\n",
        "  data_por_estudiante = pd.pivot_table(data, index= [\"student_uid\", \"cond_cant_ejer_resueltos\"], columns = [\"exercise_new_id\"], fill_value= np.nan)\n",
        "  data_por_estudiante.columns = ['_'.join(col).strip() for col in data_por_estudiante.columns.values]\n",
        "  \n",
        "  # Llenamos el último estado con 1 si no resolvió cada ejercicio\n",
        "  for col in data_por_estudiante.columns[data_por_estudiante.columns.str.contains(\"ultimo_estado\")]:\n",
        "    data_por_estudiante[col] = data_por_estudiante[col].fillna(1) \n",
        "  \n",
        "  # Imputamos los NaN's en intentos incorrectos y en tiempo invertido de cada ejercicio\n",
        "  if imputar_con_constante:\n",
        "    data_por_estudiante.fillna(imputacion_nan, inplace = True) # Asigna la imputacion_nan para variables tiempo e intentos incorrectos\n",
        "  else:\n",
        "    for col in data_por_estudiante.columns: \n",
        "      data_por_estudiante[col] = np.where(data_por_estudiante[col].isna(), data_por_estudiante[col].max(), data_por_estudiante[col])\n",
        "  return data_por_estudiante.reset_index()\n",
        "\n",
        "def regresion_logistica(df, test_size, drop_ultimo_estado):\n",
        "  X = df.drop(columns = [\"cond_cant_ejer_resueltos\"])\n",
        "  if drop_ultimo_estado:\n",
        "    X.drop(columns = X.columns[X.columns.str.contains(\"ultimo_estado\")], inplace = True)\n",
        "  y = df.cond_cant_ejer_resueltos\n",
        "  X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=test_size, random_state=0)\n",
        "  lr = LogisticRegression(random_state=0).fit(X_train, y_train)\n",
        "  return X_train, X_test, y_train, y_test, lr.predict(X_train), lr.predict(X_test)\n",
        "\n",
        "def construir_regresion_logistica(ids_ejercicios, test_size, imputar_con_constante, imputacion_nan, drop_ultimo_estado):\n",
        "  return regresion_logistica(data_por_student(ids_ejercicios, imputar_con_constante, imputacion_nan), test_size, drop_ultimo_estado)"
      ],
      "id": "yjViw0bdkAqz",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "UkxmMLx-Ug1M"
      },
      "source": [
        "# Apertura del dataset desde Google Drive\n",
        "\n",
        "from pydrive.auth import GoogleAuth\n",
        "from pydrive.drive import GoogleDrive\n",
        "from google.colab import auth\n",
        "from oauth2client.client import GoogleCredentials\n",
        "\n",
        "def connect_to_drive():\n",
        "  auth.authenticate_user()\n",
        "  gauth = GoogleAuth()\n",
        "  gauth.credentials = GoogleCredentials.get_application_default()\n",
        "  return GoogleDrive(gauth)\n",
        "\n",
        "def read_csv_from_drive(drive, id, filename):\n",
        "  downloaded = drive.CreateFile({'id':id}) \n",
        "  downloaded.GetContentFile(filename)  \n",
        "  return pd.read_csv(filename)\n",
        "\n",
        "drive = connect_to_drive()  # utilización de la función de apertura de archivos desde drive\n",
        "df_submissions_0 = read_csv_from_drive(drive, '1oqeKaq164fYieUs6WODv53tF4hl7QY9N', 'df_submissions_python.csv') # el 2° parámetro es el id del archivo (terminación del link en drive)\n",
        "\n",
        "# 1oqeKaq164fYieUs6WODv53tF4hl7QY9N\n",
        "# [Fuente: https://colab.research.google.com/drive/1bPwYa_uumNJ7N6NazqSwLr-8dmiCaIj8]\n"
      ],
      "id": "UkxmMLx-Ug1M",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "vzKtcQ9ZmDFS"
      },
      "source": [
        "# GENERACIÓN DE DF "
      ],
      "id": "vzKtcQ9ZmDFS"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "CEpmdp5Rma_V"
      },
      "source": [
        "## Limpieza y filtrado"
      ],
      "id": "CEpmdp5Rma_V"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "tGBlMiVgma_z"
      },
      "source": [
        "# Corrección de nombres\n",
        "\n",
        "df_submissions_0[\"exercise_name\"] = np.where(df_submissions_0[\"exercise_name\"] == 'Más booleanos', 'El retorno del booleano' , df_submissions_0[\"exercise_name\"])\n",
        "df_submissions_0[\"exercise_name\"] = np.where(df_submissions_0[\"exercise_name\"] == 'PIenso que así es más fácil', 'Pienso que así es más fácil', df_submissions_0[\"exercise_name\"])\n",
        "\n",
        "# Eliminamos \"Calefones\": \n",
        "df_submissions_0 = df_submissions_0[df_submissions_0[\"exercise_name\"] != \"Calefones\"]\n",
        "\n",
        "_orden_ejer = {\"No tan distintos\": 1, \"Funciones, declaración\": 2, \"Funciones, uso\": 3, \"Probando funciones\": 4, \"Haciendo cuentas\": 5, \"Poniendo topes\": 6, \"¿Es par?\": 7, \"Libros de la buena memoria\": 8, \"Booleanos\": 9, \"Palabras, sólo palabras\": 10, \"Operando strings\": 11, \"¡GRITAR!\": 12, \"¿Y qué tal si...?\":13, \"¿De qué signo sos?\": 14, \"El retorno del booleano\": 15, \"Los premios\": 16, \"Tipos de datos\": 17, \"Datos de todo tipo\": 18, \n",
        "        \"Comprando Hardware\": 19, \"¿Me conviene?\": 20, \"Triangulos\": 21, \"Cuadrados\": 22, \"¿Está afinado?\": 23, \"¿Está cerca?\": 24, \"Cartelitos\": 25, \"Más Cartelitos\": 26, \"Cartelitos óptimos\": 27, \"Cara o ceca\": 28, \"¡Envido!\": 29, \"¡Quiero retruco!\": 30, \"¡Quiero vale cuatro!\": 31,\n",
        "        \"¿Y el tablero?\": 32, \"Impresión por pantalla\": 33, \"Martin Fierro\": 34, \"¿Y los procedimientos?\": 35, \"¿Y el program?\": 36, \"Conversiones\": 37, \"El círculo de la vida\": 38, \"Pienso que así es más fácil\": 39, \"Esto no tiene valor\": 40, \"Variables globales\": 41, \"Volviéndonos ricos\": 42, \"¿Y esto cuánto vale?\": 43}\n",
        "\n",
        "# Mapeo del nombre de los ejercicios con su correspondiente orden dentro de la guía:\n",
        "df_submissions_0[\"exercise_new_id\"] = df_submissions_0[\"exercise_name\"].map(_orden_ejer)\n",
        "df_submissions_0[\"exercise_new_id\"] = df_submissions_0[\"exercise_new_id\"].apply(int)\n",
        "\n",
        "# Eliminamos la columna del viejo id ya que no sera utilizada de aqui en adelante\n",
        "df_submissions_0 = df_submissions_0.drop(\"exercise_eid\", axis = 1)\n",
        "\n",
        "# Eliminamos las columnas que no tienen datos\n",
        "df_submissions_0 = df_submissions_0.drop(\"feedback\", axis = 1)\n",
        "\n",
        "# Pasamos la fecha a un formato datetime\n",
        "df_submissions_0[\"submission_created_at\"] = pd.to_datetime(df_submissions_0['submission_created_at'])\n",
        "\n",
        "# Reemplazamos las listas vacías en test_results por NaN's\n",
        "df_submissions_0[\"test_results\"] = df_submissions_0.test_results.replace(\"[]\", np.nan)\n",
        "\n",
        "# Corregimos un outlier (failed que no compilaba y debería ser errored) a partir de result\n",
        "df_submissions_0[\"submission_status\"] = np.where((df_submissions_0.submission_status == \"failed\") & (df_submissions_0.result.notna()) , \"errored\", df_submissions_0.submission_status)\n",
        "\n",
        "# Se filtran (excluyen por el momento) las pruebas: \n",
        "df_submissions = df_submissions_0[df_submissions_0[\"sid\"] != -1]\n",
        "\n",
        "# Se filtran (excluyen por el momento) status pending y aborted:\n",
        "df_submissions = df_submissions[~df_submissions[\"submission_status\"].isin([\"pending\", \"aborted\"])]\n",
        "\n",
        "# Se filtran a estudiantes que tuvieron un único envío o son inválidos (por definición o porque su período de trabajo es mayor a un cuatrimestre):\n",
        "df_submissions = df_submissions[~df_submissions[\"student_uid\"].isin([75, 142, 361, 97, -1])]\n",
        "\n",
        "# Se filtran ejercicios que fueron modificados y resueltos, en definitiva, por muy pocos estudiantes\n",
        "df_submissions = df_submissions[df_submissions[\"exercise_new_id\"] != 7]\n",
        "\n",
        "# Obtenemos el periodo al que pertenecen l@s estudiantes para poder identificar los cursos: \n",
        "    # Creamos una columna que sólo indique el año: \n",
        "df_submissions[\"año\"] = df_submissions[\"submission_created_at\"].dt.year\n",
        "\n",
        "    # Creamos una columna que sólo indique el cuatrimestre: \n",
        "df_submissions[\"cuatrimestre\"] = np.where(df_submissions[\"submission_created_at\"].dt.month.isin([ 5,  6,  7, 8]), \"2 cuatr.\",\"3 cuatr.\")\n",
        "\n",
        "    # Creamos una columna que sólo indique el día:\n",
        "dias = {0: \"1-Lunes\", 1: \"2-Martes\", 2: \"3-Miércoles\", 3: \"4-Jueves\", 4: \"5-Viernes\", 5: \"6-Sábado\", 6: \"7-Domingo\"}\n",
        "df_submissions[\"dia\"] = df_submissions[\"submission_created_at\"].dt.dayofweek\n",
        "df_submissions[\"dia\"].replace(dias, inplace =True)\n"
      ],
      "id": "tGBlMiVgma_z",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5CJDLNT5ma_2"
      },
      "source": [
        "## Cálculo de tiempos"
      ],
      "id": "5CJDLNT5ma_2"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 424
        },
        "id": "4vMxf6Qhma_4",
        "outputId": "1665c0a9-0fd6-4029-9d84-e1a453c4e767"
      },
      "source": [
        "# Tiempo (en segundos) entre cada envío por estudiante:\n",
        "\n",
        "intervalos_tiempo_por_estudiante = df_submissions.sort_values(by=[\"student_uid\", \"submission_created_at\"]).groupby('student_uid').submission_created_at.diff().dt.total_seconds().round().reset_index()\n",
        "intervalos_tiempo_por_estudiante[\"student_uid\"] =  df_submissions.sort_values(\"student_uid\").student_uid.reset_index(drop=True)\n",
        "intervalos_tiempo_por_estudiante = intervalos_tiempo_por_estudiante.rename(columns = {\"submission_created_at\": \"tiempo_entre_envios\"}).drop(columns = \"index\")\n",
        "intervalos_tiempo_por_estudiante = intervalos_tiempo_por_estudiante[intervalos_tiempo_por_estudiante.tiempo_entre_envios.between(3, 1200)]\n",
        "\n",
        "# Tiempo total\n",
        "tiempo_recorrido_por_estudiante = intervalos_tiempo_por_estudiante.groupby(\"student_uid\").tiempo_entre_envios.sum()\n",
        "\n",
        "# Tiempo (en segundos) entre cada envío por estudiante y ejercicio:\n",
        "tiempo_entre_envios_por_ejercicio = df_submissions.sort_values(by=[\"student_uid\", \"exercise_new_id\", \"submission_created_at\"]).groupby(by=['student_uid', \"exercise_new_id\"]).submission_created_at.diff().dt.total_seconds().apply(np.ceil).reset_index()\n",
        "\n",
        "# Agregamos la información del estudiante y del ejercicio en cuestión\n",
        "tiempo_entre_envios_por_ejercicio = tiempo_entre_envios_por_ejercicio.fillna(0).rename(columns = {\"submission_created_at\": \"tiempo_entre_intentos\"}).drop(columns = [\"index\"])\n",
        "tiempo_entre_envios_por_ejercicio[[\"student_uid\", \"exercise_new_id\", \"submission_created_at\"]] = df_submissions.sort_values(by=[\"student_uid\", \"exercise_new_id\", \"submission_created_at\"]).reset_index()[[\"student_uid\", \"exercise_new_id\", \"submission_created_at\"]]\n",
        "\n",
        "tiempo_entre_envios_por_ejercicio[\"tiempo_intento_modificado\"] = tiempo_entre_envios_por_ejercicio.tiempo_entre_intentos.apply(lambda row: 0 if row > 1200 else row)\n",
        "\n",
        "# Nuevo tiempo total (en segundos) para cada estudiante en cada ejercicio\n",
        "tiempo_total_por_ejercicio_2 = tiempo_entre_envios_por_ejercicio.groupby(by=['student_uid', \"exercise_new_id\"]).agg({\"tiempo_intento_modificado\": \"sum\", \"submission_created_at\": \"nunique\"}).reset_index()\n",
        "tiempo_total_por_ejercicio_2.rename(columns = {\"tiempo_intento_modificado\": \"tiempo_total\", \"submission_created_at\": \"intentos_totales\"}, inplace = True)\n",
        "tiempo_total_por_ejercicio_2"
      ],
      "id": "4vMxf6Qhma_4",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>student_uid</th>\n",
              "      <th>exercise_new_id</th>\n",
              "      <th>tiempo_total</th>\n",
              "      <th>intentos_totales</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>76</td>\n",
              "      <td>2</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>76</td>\n",
              "      <td>3</td>\n",
              "      <td>0.0</td>\n",
              "      <td>2</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>76</td>\n",
              "      <td>6</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>76</td>\n",
              "      <td>9</td>\n",
              "      <td>915.0</td>\n",
              "      <td>5</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>76</td>\n",
              "      <td>10</td>\n",
              "      <td>20.0</td>\n",
              "      <td>2</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7810</th>\n",
              "      <td>359</td>\n",
              "      <td>43</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7811</th>\n",
              "      <td>362</td>\n",
              "      <td>21</td>\n",
              "      <td>87.0</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7812</th>\n",
              "      <td>363</td>\n",
              "      <td>25</td>\n",
              "      <td>61.0</td>\n",
              "      <td>3</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7813</th>\n",
              "      <td>363</td>\n",
              "      <td>26</td>\n",
              "      <td>151.0</td>\n",
              "      <td>4</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7814</th>\n",
              "      <td>363</td>\n",
              "      <td>27</td>\n",
              "      <td>643.0</td>\n",
              "      <td>12</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>7815 rows × 4 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "      student_uid  exercise_new_id  tiempo_total  intentos_totales\n",
              "0              76                2           0.0                 1\n",
              "1              76                3           0.0                 2\n",
              "2              76                6           0.0                 1\n",
              "3              76                9         915.0                 5\n",
              "4              76               10          20.0                 2\n",
              "...           ...              ...           ...               ...\n",
              "7810          359               43           0.0                 1\n",
              "7811          362               21          87.0                 3\n",
              "7812          363               25          61.0                 3\n",
              "7813          363               26         151.0                 4\n",
              "7814          363               27         643.0                12\n",
              "\n",
              "[7815 rows x 4 columns]"
            ]
          },
          "metadata": {},
          "execution_count": 5
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "nVRXPHjBma_8"
      },
      "source": [
        "## Construcción de nuevo df con métricas por estudiante"
      ],
      "id": "nVRXPHjBma_8"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 444
        },
        "id": "mXLaOCl6ma_-",
        "outputId": "fd394cac-431d-47f1-a891-5e9a53850dbc"
      },
      "source": [
        "# Por cada estudiante obtenemos la cantidad de ejercicios con status passed o passed_with_warnings que realizó:\n",
        "#filtro_status = submissions_correctas(df_submissions)\n",
        "filtro_status = df_submissions\n",
        "group_ejer = filtro_status[[\"student_uid\", \"exercise_new_id\"]]\n",
        "group_ejer = group_ejer.groupby([\"student_uid\"]).nunique().rename(columns = {'exercise_new_id': 'ejercicios_resueltos'})\n",
        "\n",
        "# Creamos la columna de desercion:\n",
        "# Regla de decision: \n",
        "criterio_por_ejercicios_resueltos = group_ejer[\"ejercicios_resueltos\"].quantile(0.25)\n",
        "#criterio_por_ejercicios_resueltos = 30\n",
        "\n",
        "df_cond_ejer = group_ejer.copy()\n",
        "df_cond_ejer[\"cond_cant_ejer_resueltos\"] = np.where(df_cond_ejer[\"ejercicios_resueltos\"] < criterio_por_ejercicios_resueltos, 1, 0)\n",
        "\n",
        "df_cond_ejer = df_cond_ejer.reset_index() \n",
        "\n",
        "# -------\n",
        "\n",
        "filtro_fallidos = submissions_incorrectas(df_submissions)\n",
        "\n",
        "# Envíos fallidos por estudiante, por guía/lección, y por ejercicio.\n",
        "\n",
        "df_fallidos = filtro_fallidos.groupby(['student_uid','exercise_new_id'])['sid'].count().reset_index().rename(columns = {'sid': 'intentos_fallidos_max_ejer'}).sort_values([\"student_uid\", \"intentos_fallidos_max_ejer\"], ascending = [True, False])\n",
        "\n",
        "# Regla de decisión: \n",
        "criterio_por_intentos_fallidos = df_fallidos.sort_values([\"student_uid\", \"intentos_fallidos_max_ejer\"], ascending = [True, False]).drop_duplicates(\"student_uid\")[\"intentos_fallidos_max_ejer\"].quantile(0.75)\n",
        "\n",
        "# Creamos la columna de deserción:\n",
        "df_cond_fallos_ejer = df_fallidos.copy()[[\"student_uid\", \"intentos_fallidos_max_ejer\"]]\n",
        "df_cond_fallos_ejer[\"cond_fallos_ejer\"] = np.where(df_fallidos[\"intentos_fallidos_max_ejer\"] > criterio_por_intentos_fallidos, 1, 0)\n",
        "\n",
        "df_cond_fallos_ejer = df_cond_fallos_ejer.sort_values('intentos_fallidos_max_ejer', ascending = False).drop_duplicates('student_uid')\n",
        "\n",
        "# -------\n",
        "\n",
        "# Cantidad de ejercicios intentados (en cualquier estado) por persona\n",
        "\n",
        "group_ejer_int = df_submissions[[\"student_uid\", \"exercise_new_id\"]]\n",
        "group_ejer_int = group_ejer_int.groupby([\"student_uid\"]).nunique().rename(columns = {'exercise_new_id': 'ejercicios_totales'})\n",
        "group_ejer_int = group_ejer_int.reset_index()\n",
        "\n",
        "# -------\n",
        "\n",
        "intervalos_tiempo_por_estudiante = df_submissions.sort_values(by=[\"student_uid\", \"submission_created_at\"]).groupby('student_uid').submission_created_at.diff().dt.total_seconds().round().reset_index()\n",
        "intervalos_tiempo_por_estudiante[\"student_uid\"] = df_submissions.sort_values(\"student_uid\").student_uid.reset_index(drop=True)\n",
        "intervalos_tiempo_por_estudiante = intervalos_tiempo_por_estudiante.rename(columns = {\"submission_created_at\": \"tiempo_total_invertido\"}).drop(columns = \"index\")\n",
        "\n",
        "# Nos quedamos con los intervalos de tiempo que se hayan realizado entre 3 segundos y 20 minutos\n",
        "intervalos_tiempo_por_estudiante = intervalos_tiempo_por_estudiante[intervalos_tiempo_por_estudiante.tiempo_total_invertido.between(3, 1200)]\n",
        "df_cond_tiempo_ejer = intervalos_tiempo_por_estudiante.groupby('student_uid', as_index=False).sum()\n",
        "\n",
        "\n",
        "\n",
        "# DF con las condiciones obtenidas a partir de las metricas de deserción\n",
        "\n",
        "df_condiciones = df_cond_ejer.merge(df_cond_tiempo_ejer, on = \"student_uid\", how = \"outer\")\n",
        "df_condiciones = df_condiciones.merge(df_cond_fallos_ejer, on = \"student_uid\", how = \"outer\")\n",
        "df_condiciones = df_condiciones.merge(group_ejer_int, on = \"student_uid\", how = \"outer\")\n",
        "df_condiciones = df_condiciones.fillna(0)\n",
        "df_condiciones['ejercicios_fallidos'] = df_condiciones.ejercicios_totales - df_condiciones.ejercicios_resueltos\n",
        "\n",
        "# -------\n",
        "df_condiciones['tiempo_prom_por_ej_correcto'] = df_condiciones.tiempo_total_invertido / df_condiciones.ejercicios_resueltos\n",
        "df_condiciones['tiempo_prom_por_ejercicio'] = df_condiciones.tiempo_total_invertido / df_condiciones.ejercicios_totales\n",
        "\n",
        "# Regla de decisión: \n",
        "criterio_por_tiempo_por_ejercicio = df_condiciones[\"tiempo_prom_por_ejercicio\"].quantile(0.75)\n",
        "\n",
        "df_condiciones[\"cond_tiempo_prom_ejer\"] = np.where(df_condiciones[\"tiempo_prom_por_ejercicio\"] > criterio_por_tiempo_por_ejercicio, 1, 0)\n",
        "\n",
        "# -------\n",
        "\n",
        "intentos_totales = df_submissions.groupby('student_uid', as_index=False).sid.count().rename(columns={\"sid\":\"intentos_totales\"}) # envios totales por persona\n",
        "intentos_incorrectos = submissions_incorrectas(df_submissions).groupby('student_uid', as_index=False).sid.count().rename(columns={\"sid\":\"intentos_incorrectos\"}) # envios fallidos por persona\n",
        "intentos_correctos = submissions_correctas(df_submissions).groupby('student_uid', as_index=False).sid.count().rename(columns={\"sid\":\"intentos_correctos\"}) # envios correctos por persona\n",
        "\n",
        "# Agregamos las columnas con cantidad de intentos totales y distinguidos por estado\n",
        "df_condiciones = df_condiciones.merge(intentos_totales, on = \"student_uid\", how = \"outer\")\n",
        "df_condiciones = df_condiciones.merge(intentos_incorrectos, on = \"student_uid\", how = \"outer\")\n",
        "df_condiciones = df_condiciones.merge(intentos_correctos, on = \"student_uid\", how = \"outer\")\n",
        "\n",
        "df_condiciones = df_condiciones[[\"student_uid\", \"ejercicios_totales\", \"ejercicios_fallidos\",\"ejercicios_resueltos\", \"cond_cant_ejer_resueltos\", \"intentos_totales\",\"intentos_incorrectos\",\"intentos_correctos\", \"intentos_fallidos_max_ejer\",\"cond_fallos_ejer\",\"tiempo_total_invertido\",\"tiempo_prom_por_ej_correcto\",\"tiempo_prom_por_ejercicio\",\"cond_tiempo_prom_ejer\"]]\n",
        "df_condiciones"
      ],
      "id": "mXLaOCl6ma_-",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>student_uid</th>\n",
              "      <th>ejercicios_totales</th>\n",
              "      <th>ejercicios_fallidos</th>\n",
              "      <th>ejercicios_resueltos</th>\n",
              "      <th>cond_cant_ejer_resueltos</th>\n",
              "      <th>intentos_totales</th>\n",
              "      <th>intentos_incorrectos</th>\n",
              "      <th>intentos_correctos</th>\n",
              "      <th>intentos_fallidos_max_ejer</th>\n",
              "      <th>cond_fallos_ejer</th>\n",
              "      <th>tiempo_total_invertido</th>\n",
              "      <th>tiempo_prom_por_ej_correcto</th>\n",
              "      <th>tiempo_prom_por_ejercicio</th>\n",
              "      <th>cond_tiempo_prom_ejer</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>76</td>\n",
              "      <td>30</td>\n",
              "      <td>0</td>\n",
              "      <td>30</td>\n",
              "      <td>0</td>\n",
              "      <td>60</td>\n",
              "      <td>29.0</td>\n",
              "      <td>31.0</td>\n",
              "      <td>7.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>7991.0</td>\n",
              "      <td>266.366667</td>\n",
              "      <td>266.366667</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>77</td>\n",
              "      <td>33</td>\n",
              "      <td>0</td>\n",
              "      <td>33</td>\n",
              "      <td>0</td>\n",
              "      <td>122</td>\n",
              "      <td>85.0</td>\n",
              "      <td>37.0</td>\n",
              "      <td>30.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>10945.0</td>\n",
              "      <td>331.666667</td>\n",
              "      <td>331.666667</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>78</td>\n",
              "      <td>5</td>\n",
              "      <td>0</td>\n",
              "      <td>5</td>\n",
              "      <td>1</td>\n",
              "      <td>28</td>\n",
              "      <td>25.0</td>\n",
              "      <td>3.0</td>\n",
              "      <td>11.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>3247.0</td>\n",
              "      <td>649.400000</td>\n",
              "      <td>649.400000</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>79</td>\n",
              "      <td>18</td>\n",
              "      <td>0</td>\n",
              "      <td>18</td>\n",
              "      <td>1</td>\n",
              "      <td>64</td>\n",
              "      <td>47.0</td>\n",
              "      <td>17.0</td>\n",
              "      <td>10.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>4464.0</td>\n",
              "      <td>248.000000</td>\n",
              "      <td>248.000000</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>80</td>\n",
              "      <td>13</td>\n",
              "      <td>0</td>\n",
              "      <td>13</td>\n",
              "      <td>1</td>\n",
              "      <td>91</td>\n",
              "      <td>80.0</td>\n",
              "      <td>11.0</td>\n",
              "      <td>26.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>7091.0</td>\n",
              "      <td>545.461538</td>\n",
              "      <td>545.461538</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>...</th>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "      <td>...</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>276</th>\n",
              "      <td>355</td>\n",
              "      <td>8</td>\n",
              "      <td>0</td>\n",
              "      <td>8</td>\n",
              "      <td>1</td>\n",
              "      <td>15</td>\n",
              "      <td>11.0</td>\n",
              "      <td>4.0</td>\n",
              "      <td>4.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>2020.0</td>\n",
              "      <td>252.500000</td>\n",
              "      <td>252.500000</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>277</th>\n",
              "      <td>358</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>6</td>\n",
              "      <td>5.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>5.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>96.0</td>\n",
              "      <td>96.000000</td>\n",
              "      <td>96.000000</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>278</th>\n",
              "      <td>359</td>\n",
              "      <td>13</td>\n",
              "      <td>0</td>\n",
              "      <td>13</td>\n",
              "      <td>1</td>\n",
              "      <td>31</td>\n",
              "      <td>19.0</td>\n",
              "      <td>12.0</td>\n",
              "      <td>6.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>4322.0</td>\n",
              "      <td>332.461538</td>\n",
              "      <td>332.461538</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>279</th>\n",
              "      <td>362</td>\n",
              "      <td>1</td>\n",
              "      <td>0</td>\n",
              "      <td>1</td>\n",
              "      <td>1</td>\n",
              "      <td>3</td>\n",
              "      <td>3.0</td>\n",
              "      <td>NaN</td>\n",
              "      <td>3.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>87.0</td>\n",
              "      <td>87.000000</td>\n",
              "      <td>87.000000</td>\n",
              "      <td>0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>280</th>\n",
              "      <td>363</td>\n",
              "      <td>3</td>\n",
              "      <td>0</td>\n",
              "      <td>3</td>\n",
              "      <td>1</td>\n",
              "      <td>19</td>\n",
              "      <td>18.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>12.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>1753.0</td>\n",
              "      <td>584.333333</td>\n",
              "      <td>584.333333</td>\n",
              "      <td>1</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "<p>281 rows × 14 columns</p>\n",
              "</div>"
            ],
            "text/plain": [
              "     student_uid  ...  cond_tiempo_prom_ejer\n",
              "0             76  ...                      0\n",
              "1             77  ...                      0\n",
              "2             78  ...                      1\n",
              "3             79  ...                      0\n",
              "4             80  ...                      0\n",
              "..           ...  ...                    ...\n",
              "276          355  ...                      0\n",
              "277          358  ...                      0\n",
              "278          359  ...                      0\n",
              "279          362  ...                      0\n",
              "280          363  ...                      1\n",
              "\n",
              "[281 rows x 14 columns]"
            ]
          },
          "metadata": {},
          "execution_count": 6
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "jEfV3Mq2mbAB"
      },
      "source": [
        "## Construcción de nuevo df con métricas por estudiante y ejercicio"
      ],
      "id": "jEfV3Mq2mbAB"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 313
        },
        "id": "TSFKV9_XmbAC",
        "outputId": "78638185-c6c0-4b9b-bd6e-849ae48f8545"
      },
      "source": [
        "ultimo_estado = ordenar_y_agrupar(df_submissions).tail(1)[[\"student_uid\", \"exercise_new_id\", \"submission_status\"]]\n",
        "ultimo_estado = ultimo_estado.rename(columns={\"submission_status\": \"ultimo_estado\"})\n",
        "\n",
        "#Tablas por estudiante y ejercicio con cantidad de intentos en cada estado:\n",
        "correctos = ordenar_y_agrupar(submissions_correctas(df_submissions)).agg({\"sid\": \"count\", \"submission_created_at\": \"min\"})\n",
        "correctos = correctos.rename(columns = {\"sid\": \"intentos_correctos\", \"submission_created_at\": \"fecha_primer_correcto\"})\n",
        "\n",
        "incorrectos = ordenar_y_agrupar(submissions_incorrectas(df_submissions)).agg({\"sid\": \"count\", \"submission_created_at\": \"max\"})\n",
        "incorrectos = incorrectos.rename(columns = {\"sid\": \"intentos_incorrectos\", \"submission_created_at\": \"fecha_ultimo_incorrecto\"})\n",
        "\n",
        "estudiante_por_ejercicio = pd.concat([correctos, incorrectos], axis = 1).reset_index()\n",
        "\n",
        "#-----\n",
        "\n",
        "estudiante_por_ejercicio = pd.merge(estudiante_por_ejercicio, ultimo_estado)\n",
        "estudiante_por_ejercicio = pd.merge(estudiante_por_ejercicio, tiempo_total_por_ejercicio_2)\n",
        "\n",
        "#-----\n",
        "\n",
        "# Reordenamos las columnas\n",
        "estudiante_por_ejercicio = estudiante_por_ejercicio[[\"student_uid\", \"exercise_new_id\", \"ultimo_estado\", \"intentos_totales\", \"intentos_correctos\", \"intentos_incorrectos\", \"tiempo_total\", \"fecha_primer_correcto\", \"fecha_ultimo_incorrecto\"]]\n",
        "estudiante_por_ejercicio\n",
        "\n",
        "#-----\n",
        "\n",
        "# Agregamos columnas de condiciones para tener más info y distinguir deserción\n",
        "estudiante_por_ejercicio = pd.merge(estudiante_por_ejercicio, df_condiciones[[\"student_uid\", \"ejercicios_totales\",\"ejercicios_resueltos\", \"cond_cant_ejer_resueltos\", \"intentos_fallidos_max_ejer\"]], on = \"student_uid\", how = \"left\")\n",
        "estudiante_por_ejercicio.fillna(0, inplace = True)\n",
        "\n",
        "estudiante_por_ejercicio.head()\n"
      ],
      "id": "TSFKV9_XmbAC",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>student_uid</th>\n",
              "      <th>exercise_new_id</th>\n",
              "      <th>ultimo_estado</th>\n",
              "      <th>intentos_totales</th>\n",
              "      <th>intentos_correctos</th>\n",
              "      <th>intentos_incorrectos</th>\n",
              "      <th>tiempo_total</th>\n",
              "      <th>fecha_primer_correcto</th>\n",
              "      <th>fecha_ultimo_incorrecto</th>\n",
              "      <th>ejercicios_totales</th>\n",
              "      <th>ejercicios_resueltos</th>\n",
              "      <th>cond_cant_ejer_resueltos</th>\n",
              "      <th>intentos_fallidos_max_ejer</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>76</td>\n",
              "      <td>2</td>\n",
              "      <td>passed</td>\n",
              "      <td>1</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>2019-08-30 13:17:17.700000</td>\n",
              "      <td>0</td>\n",
              "      <td>30</td>\n",
              "      <td>30</td>\n",
              "      <td>0</td>\n",
              "      <td>7.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>76</td>\n",
              "      <td>3</td>\n",
              "      <td>passed</td>\n",
              "      <td>2</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>2019-09-10 20:00:32.156000</td>\n",
              "      <td>2019-08-30 13:40:26.023000</td>\n",
              "      <td>30</td>\n",
              "      <td>30</td>\n",
              "      <td>0</td>\n",
              "      <td>7.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>76</td>\n",
              "      <td>6</td>\n",
              "      <td>passed</td>\n",
              "      <td>1</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>2019-09-23 12:24:41.746000</td>\n",
              "      <td>0</td>\n",
              "      <td>30</td>\n",
              "      <td>30</td>\n",
              "      <td>0</td>\n",
              "      <td>7.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>76</td>\n",
              "      <td>9</td>\n",
              "      <td>passed</td>\n",
              "      <td>5</td>\n",
              "      <td>1.0</td>\n",
              "      <td>4.0</td>\n",
              "      <td>915.0</td>\n",
              "      <td>2019-09-23 13:05:45.340000</td>\n",
              "      <td>2019-09-23 13:04:41.529000</td>\n",
              "      <td>30</td>\n",
              "      <td>30</td>\n",
              "      <td>0</td>\n",
              "      <td>7.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>76</td>\n",
              "      <td>10</td>\n",
              "      <td>passed</td>\n",
              "      <td>2</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>20.0</td>\n",
              "      <td>2019-09-23 13:08:05.815000</td>\n",
              "      <td>2019-09-23 13:07:46.418000</td>\n",
              "      <td>30</td>\n",
              "      <td>30</td>\n",
              "      <td>0</td>\n",
              "      <td>7.0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   student_uid  ...  intentos_fallidos_max_ejer\n",
              "0           76  ...                         7.0\n",
              "1           76  ...                         7.0\n",
              "2           76  ...                         7.0\n",
              "3           76  ...                         7.0\n",
              "4           76  ...                         7.0\n",
              "\n",
              "[5 rows x 13 columns]"
            ]
          },
          "metadata": {},
          "execution_count": 7
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "54jOL7Z1ImKs"
      },
      "source": [
        "\n",
        "\n",
        "# 1 - DF CON INFORMACIÓN DE CADA EJERCICIO POR ESTUDIANTE"
      ],
      "id": "54jOL7Z1ImKs"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-DsywqfD2UYs"
      },
      "source": [
        "Generamos un nuevo dataframe con las mismas métricas que teníamos, pero ahora \"por estudiante y por ejercicio\".\n",
        "\n",
        "Es decir, tendremos variables que indiquen:\n",
        "- el tiempo invertido por cada ejercicio, y\n",
        "- la cantidad de intentos fallidos para cada ejercicio."
      ],
      "id": "-DsywqfD2UYs"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 313
        },
        "id": "YDfaJpJUIhWU",
        "cellView": "form",
        "outputId": "a4012ede-351d-4019-ec90-be70ba95fd48"
      },
      "source": [
        "#@title\n",
        "estudiante_por_ejercicio.head()"
      ],
      "id": "YDfaJpJUIhWU",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>student_uid</th>\n",
              "      <th>exercise_new_id</th>\n",
              "      <th>ultimo_estado</th>\n",
              "      <th>intentos_totales</th>\n",
              "      <th>intentos_correctos</th>\n",
              "      <th>intentos_incorrectos</th>\n",
              "      <th>tiempo_total</th>\n",
              "      <th>fecha_primer_correcto</th>\n",
              "      <th>fecha_ultimo_incorrecto</th>\n",
              "      <th>ejercicios_totales</th>\n",
              "      <th>ejercicios_resueltos</th>\n",
              "      <th>cond_cant_ejer_resueltos</th>\n",
              "      <th>intentos_fallidos_max_ejer</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>0</th>\n",
              "      <td>76</td>\n",
              "      <td>2</td>\n",
              "      <td>passed</td>\n",
              "      <td>1</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>2019-08-30 13:17:17.700000</td>\n",
              "      <td>0</td>\n",
              "      <td>30</td>\n",
              "      <td>30</td>\n",
              "      <td>0</td>\n",
              "      <td>7.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>1</th>\n",
              "      <td>76</td>\n",
              "      <td>3</td>\n",
              "      <td>passed</td>\n",
              "      <td>2</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>2019-09-10 20:00:32.156000</td>\n",
              "      <td>2019-08-30 13:40:26.023000</td>\n",
              "      <td>30</td>\n",
              "      <td>30</td>\n",
              "      <td>0</td>\n",
              "      <td>7.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>2</th>\n",
              "      <td>76</td>\n",
              "      <td>6</td>\n",
              "      <td>passed</td>\n",
              "      <td>1</td>\n",
              "      <td>1.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>0.0</td>\n",
              "      <td>2019-09-23 12:24:41.746000</td>\n",
              "      <td>0</td>\n",
              "      <td>30</td>\n",
              "      <td>30</td>\n",
              "      <td>0</td>\n",
              "      <td>7.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>3</th>\n",
              "      <td>76</td>\n",
              "      <td>9</td>\n",
              "      <td>passed</td>\n",
              "      <td>5</td>\n",
              "      <td>1.0</td>\n",
              "      <td>4.0</td>\n",
              "      <td>915.0</td>\n",
              "      <td>2019-09-23 13:05:45.340000</td>\n",
              "      <td>2019-09-23 13:04:41.529000</td>\n",
              "      <td>30</td>\n",
              "      <td>30</td>\n",
              "      <td>0</td>\n",
              "      <td>7.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>4</th>\n",
              "      <td>76</td>\n",
              "      <td>10</td>\n",
              "      <td>passed</td>\n",
              "      <td>2</td>\n",
              "      <td>1.0</td>\n",
              "      <td>1.0</td>\n",
              "      <td>20.0</td>\n",
              "      <td>2019-09-23 13:08:05.815000</td>\n",
              "      <td>2019-09-23 13:07:46.418000</td>\n",
              "      <td>30</td>\n",
              "      <td>30</td>\n",
              "      <td>0</td>\n",
              "      <td>7.0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "   student_uid  ...  intentos_fallidos_max_ejer\n",
              "0           76  ...                         7.0\n",
              "1           76  ...                         7.0\n",
              "2           76  ...                         7.0\n",
              "3           76  ...                         7.0\n",
              "4           76  ...                         7.0\n",
              "\n",
              "[5 rows x 13 columns]"
            ]
          },
          "metadata": {},
          "execution_count": 35
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "PdOC28Ni9vjW"
      },
      "source": [
        "# 2- ANALISIS PARA DETECTAR EJERCICIOS DIFÍCILES\n",
        "\n",
        "Objetivo: Seleccionar aquellas variables \"con información por ejercicio\", que den mejores predicciones de estudiantes desertores.\n"
      ],
      "id": "PdOC28Ni9vjW"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "hbgD4Sge0Etk"
      },
      "source": [
        "### Análisis 1- Cuánta gente llegó a cada ejercicio como máximo"
      ],
      "id": "hbgD4Sge0Etk"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "oTFAcGpOxxai"
      },
      "source": [
        "En el gráfico y tabla que se encuentran a continuación, vemos cuánta gente llegó a cada ejercicio como máximo y no pudo continuar resolviendo satisfactoriamente los siguientes. "
      ],
      "id": "oTFAcGpOxxai"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ffNszKkT9vjX",
        "cellView": "form"
      },
      "source": [
        "#@title\n",
        "# Para ver la cantidad de estudiantes que llegaron como máximo, a cada ejercicio.\n",
        "\n",
        "ejercicios_alcanzados = estudiante_por_ejercicio.groupby(\"student_uid\").exercise_new_id.max().value_counts().reset_index()\n",
        "ejercicios_alcanzados.rename(columns= {\"exercise_new_id\": \"estudiantes\", \"index\": \"exercise_new_id\"}, inplace = True)"
      ],
      "id": "ffNszKkT9vjX",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "KRn-cbo1nh9e",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 357
        },
        "cellView": "form",
        "outputId": "2c8404f8-0f27-44c4-e2f7-d3de420d70f4"
      },
      "source": [
        "#@title\n",
        "ejercicios_alcanzados.sort_values(\"exercise_new_id\").plot.bar(x = \"exercise_new_id\", y = \"estudiantes\", figsize = (10, 5) )"
      ],
      "id": "KRn-cbo1nh9e",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7fc45cc57390>"
            ]
          },
          "metadata": {},
          "execution_count": 40
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x360 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "hpk7FicTPIEe",
        "outputId": "9df58c9a-4eb6-4f7c-ea39-4a402ae16e69"
      },
      "source": [
        "estudiante_por_ejercicio.groupby(\"student_uid\").exercise_new_id.max().value_counts().head(10)"
      ],
      "id": "hpk7FicTPIEe",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "43    198\n",
              "9       8\n",
              "3       6\n",
              "10      6\n",
              "37      5\n",
              "27      5\n",
              "20      4\n",
              "6       4\n",
              "11      4\n",
              "12      4\n",
              "Name: exercise_new_id, dtype: int64"
            ]
          },
          "metadata": {},
          "execution_count": 39
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "rCcXJmRx_-Cz"
      },
      "source": [
        "* Sin considerar el n° 43 (final del recorrido), los picos indicarían aquellos ejercicios \"difíciles\" que dejaron a más gente en el camino.\n",
        "\n",
        "* Recordar que no están todos los ejercicios listados, ya que fueron descartados los de consola."
      ],
      "id": "rCcXJmRx_-Cz"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "PLoBwPsE0vUg"
      },
      "source": [
        "### Análisis 2- Relación entre resoluciones correctas e intentos fallidos, promedio por persona [para cada ejercicio]."
      ],
      "id": "PLoBwPsE0vUg"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "XrkdP-7H9vjZ",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 300
        },
        "outputId": "6bd21bd6-1b19-40c6-9a07-556085afa4f9"
      },
      "source": [
        "# Gráfico de dispersión de los ejercicios, considerando para cada ejercicio el par ordenado:\n",
        "#  cantidad de intentos incorrectos promedio por ejercicio \n",
        "#  y cantidad de resoluciones correctas (ya sean passed o passed with warning) por ejercicio\n",
        "\n",
        "info_ejercicios = pd.merge(estudiante_por_ejercicio.groupby(\"exercise_new_id\").agg({\"intentos_incorrectos\":'sum', \"student_uid\": \"nunique\"}).reset_index().rename(columns={\"student_uid\": \"estudiantes\"}), estudiante_por_ejercicio[(estudiante_por_ejercicio.ultimo_estado == \"passed\") | (estudiante_por_ejercicio.ultimo_estado == \"passed_with_warnings\")].groupby('exercise_new_id').student_uid.nunique().reset_index().rename(columns={\"student_uid\": \"resoluciones_correctas\"}))\n",
        "info_ejercicios[\"incorrectos_promedio\"] = info_ejercicios.intentos_incorrectos / info_ejercicios.estudiantes\n",
        "info_ejercicios.plot.scatter(y = \"resoluciones_correctas\", x = \"incorrectos_promedio\")"
      ],
      "id": "XrkdP-7H9vjZ",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7fc45f6f21d0>"
            ]
          },
          "metadata": {},
          "execution_count": 10
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 432x288 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "ahuQU1y319Sl"
      },
      "source": [
        "En el gráfico se destacan a la derecha cuatro ejercicios (n° 10, 11, 20 y 26), con gran cantidad de intentos incorrectos (mayor a 7)."
      ],
      "id": "ahuQU1y319Sl"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "qe53TLAS9vja",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 175
        },
        "outputId": "75d163f8-398c-4463-c529-ed4abbee1bed"
      },
      "source": [
        "# Para identificar los cuatro ejercicio de la derecha del gráfico\n",
        "\n",
        "info_ejercicios[info_ejercicios.incorrectos_promedio > 7]"
      ],
      "id": "qe53TLAS9vja",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>exercise_new_id</th>\n",
              "      <th>intentos_incorrectos</th>\n",
              "      <th>estudiantes</th>\n",
              "      <th>resoluciones_correctas</th>\n",
              "      <th>incorrectos_promedio</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>10</td>\n",
              "      <td>2350.0</td>\n",
              "      <td>248</td>\n",
              "      <td>228</td>\n",
              "      <td>9.475806</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>7</th>\n",
              "      <td>11</td>\n",
              "      <td>1911.0</td>\n",
              "      <td>240</td>\n",
              "      <td>230</td>\n",
              "      <td>7.962500</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>15</th>\n",
              "      <td>20</td>\n",
              "      <td>1845.0</td>\n",
              "      <td>217</td>\n",
              "      <td>192</td>\n",
              "      <td>8.502304</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21</th>\n",
              "      <td>26</td>\n",
              "      <td>1742.0</td>\n",
              "      <td>215</td>\n",
              "      <td>199</td>\n",
              "      <td>8.102326</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "    exercise_new_id  ...  incorrectos_promedio\n",
              "6                10  ...              9.475806\n",
              "7                11  ...              7.962500\n",
              "15               20  ...              8.502304\n",
              "21               26  ...              8.102326\n",
              "\n",
              "[4 rows x 5 columns]"
            ]
          },
          "metadata": {},
          "execution_count": 11
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "eo4NLjRK0NnL"
      },
      "source": [
        "### Análisis 3- Gente que no pasó cada ejercicio"
      ],
      "id": "eo4NLjRK0NnL"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "hQq44w_E1ihu"
      },
      "source": [
        "Se obtiene como la diferencia entre la gente que intentó resolver cada ejercicio, y los que lo resolvieron correctamente."
      ],
      "id": "hQq44w_E1ihu"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zd6E6E719vjc",
        "cellView": "form"
      },
      "source": [
        "#@title\n",
        "info_ejercicios[\"gente_que_no_paso\"] = info_ejercicios.estudiantes - info_ejercicios.resoluciones_correctas"
      ],
      "id": "zd6E6E719vjc",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "6Ap5mAOP9vjd",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 226
        },
        "cellView": "form",
        "outputId": "5bcdc610-8802-4952-ffbf-e4b1aad5a841"
      },
      "source": [
        "#@title\n",
        "# De acá podemos sacar los ejercicios con más deserción\n",
        "\n",
        "info_ejercicios.sort_values(\"gente_que_no_paso\", ascending = False).head()"
      ],
      "id": "6Ap5mAOP9vjd",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>exercise_new_id</th>\n",
              "      <th>intentos_incorrectos</th>\n",
              "      <th>estudiantes</th>\n",
              "      <th>resoluciones_correctas</th>\n",
              "      <th>incorrectos_promedio</th>\n",
              "      <th>gente_que_no_paso</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>15</th>\n",
              "      <td>20</td>\n",
              "      <td>1845.0</td>\n",
              "      <td>217</td>\n",
              "      <td>192</td>\n",
              "      <td>8.502304</td>\n",
              "      <td>25</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>6</th>\n",
              "      <td>10</td>\n",
              "      <td>2350.0</td>\n",
              "      <td>248</td>\n",
              "      <td>228</td>\n",
              "      <td>9.475806</td>\n",
              "      <td>20</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>5</th>\n",
              "      <td>9</td>\n",
              "      <td>1019.0</td>\n",
              "      <td>260</td>\n",
              "      <td>241</td>\n",
              "      <td>3.919231</td>\n",
              "      <td>19</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>22</th>\n",
              "      <td>27</td>\n",
              "      <td>1309.0</td>\n",
              "      <td>205</td>\n",
              "      <td>186</td>\n",
              "      <td>6.385366</td>\n",
              "      <td>19</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>21</th>\n",
              "      <td>26</td>\n",
              "      <td>1742.0</td>\n",
              "      <td>215</td>\n",
              "      <td>199</td>\n",
              "      <td>8.102326</td>\n",
              "      <td>16</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "    exercise_new_id  ...  gente_que_no_paso\n",
              "15               20  ...                 25\n",
              "6                10  ...                 20\n",
              "5                 9  ...                 19\n",
              "22               27  ...                 19\n",
              "21               26  ...                 16\n",
              "\n",
              "[5 rows x 6 columns]"
            ]
          },
          "metadata": {},
          "execution_count": 36
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7BN3L7TXlzq1"
      },
      "source": [
        "# 3- ANALISIS DE LA DESERCIÓN SÓLO EN LOS EJERCICIOS DIFÍCILES.\n",
        "\n"
      ],
      "id": "7BN3L7TXlzq1"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "sjgzNjbf5kq-"
      },
      "source": [
        "- En los gráficos siguientes, cada punto representa un estudiante, diferenciando con color a los desertores (rojo) de los no desertores (azules).\n",
        "\n",
        "- La variable \"proxy\" de deserción es la métrica cantidad de ejercicios resueltos (en color). Es decir, de las tres métricas que veníamos manejando: intentos fallidos, tiempo invertido en la resolución y cantidad de ejercicios resueltos, elegimos esta última como indicadora o métrica de deserción.\n",
        "\n",
        "- Las otras dos métricas: intentos incorrectos, y tiempo total invertido por estudiante, las mostramos como variables X e Y en los scatterplot.\n",
        "\n",
        "- **Pretendíamos ver si había una distición clara entre ambos grupos de estudiantes (desertores y no), de acuerdo a las variables en cuestión.**\n",
        "\n"
      ],
      "id": "sjgzNjbf5kq-"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Bqz48PO0pmpk",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 628
        },
        "outputId": "ecbddbb3-22c1-4296-c7b1-b910d00d920c"
      },
      "source": [
        "plot_scatter_grid('intentos_incorrectos', 'tiempo_total', [3, 9, 10, 11, 20])"
      ],
      "id": "Bqz48PO0pmpk",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 1440x1080 with 5 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ZMREsVM8y1fo",
        "cellView": "form"
      },
      "source": [
        "#@title\n",
        "# Base con desagregación de ejercicios, por variables: intentos_incorrectos y tiempo_total.\n",
        "\n",
        "data = estudiante_por_ejercicio[estudiante_por_ejercicio.exercise_new_id.isin([1,9,10,20,3,11])][[\"student_uid\", \"cond_cant_ejer_resueltos\", \"exercise_new_id\", \"intentos_incorrectos\", \"tiempo_total\"]]\n",
        "data[\"exercise_new_id\"] = data[\"exercise_new_id\"].astype(str)\n",
        "data_por_estudiante = pd.pivot_table(data, index= [\"student_uid\", \"cond_cant_ejer_resueltos\"], columns = [\"exercise_new_id\"], fill_value=0)\n",
        "\n",
        "data_por_estudiante.columns = ['_'.join(col).strip() for col in data_por_estudiante.columns.values]\n",
        "\n",
        "data_por_estudiante = data_por_estudiante.reset_index()"
      ],
      "id": "ZMREsVM8y1fo",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mS7rWQ_S7p2b",
        "cellView": "form"
      },
      "source": [
        "#@title\n",
        "data_por_estudiante.head()"
      ],
      "id": "mS7rWQ_S7p2b",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5opwYz9dwYvD"
      },
      "source": [
        "* **Conclusión: En general no vemos \"gráficamente\" una \"división\" clara entre desertores y no desertores, en los ejercicios considerados \"difíciles\".**\n",
        "\n",
        "* Notar: Dentro de los análisis de ejercicios difíciles, había aparecido el n° 7. Sin embargo, no se encuentra en estos gráficos ni tablas, ya que advertimos en el proceso que tenía menos observaciones que los demás (relacionado con el cambio de nombre visto en la entrega 1). Decidimos quitarlo del análisis.\n",
        "\n",
        "* **Próximo paso**: procedemos a **generar un modelo logísitco utilizando como *features* la combinación entre las variables tiempo invertido e intentos incorrectos, con cada uno de los ejercicios seleccionados como DIFÍCILES**, para ver si de esta manera obtenemos una buena predicción de la deserción. \n",
        "\n",
        "* Este modelo servirá como \"*baseline*\" de comparación frente a otras alternativas posteriores."
      ],
      "id": "5opwYz9dwYvD"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "HZHvFlsfq-vt"
      },
      "source": [
        "# 4 - FUNCIÓN LOGÍSTICA (BASELINE) PARA PREDECIR DESERCIÓN"
      ],
      "id": "HZHvFlsfq-vt"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "NGgNFPEW-tgP"
      },
      "source": [
        "- Pensamos que podía ser apropiado el Modelo logístico para predecir cuál estudiante deserta, y cuál no.\n",
        "\n",
        "- Hicimos un primer aprendizaje “exploratorio” o modelo baseline utilizando sólo los ejercicios difíciles: 3, 9, 10, 11, y 20."
      ],
      "id": "NGgNFPEW-tgP"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "svu8URhlrN6k",
        "cellView": "form"
      },
      "source": [
        "#@title\n",
        "X = data_por_estudiante.drop(columns = [\"cond_cant_ejer_resueltos\"])\n",
        "y = data_por_estudiante.cond_cant_ejer_resueltos\n",
        "\n",
        "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=0)"
      ],
      "id": "svu8URhlrN6k",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "zkPmRygfrbre",
        "cellView": "form"
      },
      "source": [
        "#@title\n",
        "lr = LogisticRegression(random_state=0).fit(X_train, y_train)\n",
        "y_train_pred_lr = lr.predict(X_train)\n",
        "y_test_pred_lr = lr.predict(X_test)"
      ],
      "id": "zkPmRygfrbre",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ojAgIsr2rbri",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "e2221ca7-38ba-450f-e79d-4b8d4f70845c"
      },
      "source": [
        "print(classification_report(y_train, y_train_pred_lr))"
      ],
      "id": "ojAgIsr2rbri",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.81      0.92      0.86       146\n",
            "           1       0.56      0.32      0.41        47\n",
            "\n",
            "    accuracy                           0.77       193\n",
            "   macro avg       0.68      0.62      0.63       193\n",
            "weighted avg       0.75      0.77      0.75       193\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "g87WEwM8rbrl",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "cdd3722e-a935-4efc-acd8-9a21248684c0"
      },
      "source": [
        "print(classification_report(y_test, y_test_pred_lr))"
      ],
      "id": "g87WEwM8rbrl",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.86      0.97      0.91        65\n",
            "           1       0.82      0.47      0.60        19\n",
            "\n",
            "    accuracy                           0.86        84\n",
            "   macro avg       0.84      0.72      0.76        84\n",
            "weighted avg       0.85      0.86      0.84        84\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "QEEw1O90nirY"
      },
      "source": [
        "- Como se puede apreciar, no son buenas las métricas de performance obtenidas (ver recall de clase 1 = 0.47)\n"
      ],
      "id": "QEEw1O90nirY"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "BkK68BiRoPz_"
      },
      "source": [
        "## Regresión Logística con SMOTE (prueba)"
      ],
      "id": "BkK68BiRoPz_"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "-atvspO-sqx9"
      },
      "source": [
        "En los resultados del modelo baseline observamos un buen accuracy tanto en train como en test.\n",
        "\n",
        "Sin embargo, obtuvimos un recall bajo en ambos conjuntos para la clase 1, lo que indicaría que el modelo está prediciendo o clasificando sólo el 47% de los alumnos desertores ante la presencia de \"nuevos datos\" (ver en resultados de test).\n",
        "Esto podría ser un problema relacionado con el desbalanceo de clases, donde la 1 (desertores) está subrepresentada:"
      ],
      "id": "-atvspO-sqx9"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "VMa9cyLx8RnZ"
      },
      "source": [
        "data_por_estudiante.cond_cant_ejer_resueltos.value_counts()"
      ],
      "id": "VMa9cyLx8RnZ",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gtpE9W0MtYGm"
      },
      "source": [
        "Considerando esta situación, decidimos utilizar la técnica SMOTE (Synthetic Minority Over-sampling Technique) para abordar el desbalanceo."
      ],
      "id": "gtpE9W0MtYGm"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4h1etZr77wlZ",
        "cellView": "form"
      },
      "source": [
        "#@title\n",
        "X = data_por_estudiante.drop(columns = [\"cond_cant_ejer_resueltos\"])\n",
        "y = data_por_estudiante[[\"cond_cant_ejer_resueltos\"]]\n",
        "\n",
        "oversample = SMOTE(random_state=0)\n",
        "X_smo, y_smo = oversample.fit_resample(X, y)\n",
        "\n",
        "X_train_smo, X_test_smo, y_train_smo, y_test_smo = train_test_split(X_smo, y_smo, test_size=0.3, random_state=0)"
      ],
      "id": "4h1etZr77wlZ",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xqnGCZog7nd7",
        "cellView": "form"
      },
      "source": [
        "#@title\n",
        "lr_smo = LogisticRegression(random_state=0).fit(X_train_smo, y_train_smo)\n",
        "y_train_pred_lr_smo = lr_smo.predict(X_train_smo)\n",
        "y_test_pred_lr_smo = lr_smo.predict(X_test_smo)"
      ],
      "id": "xqnGCZog7nd7",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Y1iKYT0LAJZw",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "38eb268c-5b19-4ca1-99ee-9811c640b1a0"
      },
      "source": [
        "print(classification_report(y_train_smo, y_train_pred_lr_smo))"
      ],
      "id": "Y1iKYT0LAJZw",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.87      0.76      0.81       144\n",
            "           1       0.80      0.89      0.84       151\n",
            "\n",
            "    accuracy                           0.83       295\n",
            "   macro avg       0.83      0.83      0.83       295\n",
            "weighted avg       0.83      0.83      0.83       295\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "9wP3kpxTqTC-",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "outputId": "26742bf0-85b1-41bb-ea0f-7e3f724de85f"
      },
      "source": [
        "print(classification_report(y_test_smo, y_test_pred_lr_smo))"
      ],
      "id": "9wP3kpxTqTC-",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.78      0.75      0.76        67\n",
            "           1       0.73      0.77      0.75        60\n",
            "\n",
            "    accuracy                           0.76       127\n",
            "   macro avg       0.76      0.76      0.76       127\n",
            "weighted avg       0.76      0.76      0.76       127\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "cUtjyvd1BWMZ"
      },
      "source": [
        "# 5- VARIACIONES AL MODELO BASELINE"
      ],
      "id": "cUtjyvd1BWMZ"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "9DhktGIjBKjr"
      },
      "source": [
        "PRUEBA CON SMOTE: \n",
        "\n",
        "- Si bien con SMOTE, mejoramos las métricas de performance del modelo baseline, decidimos no utilizarlo en los próximos modelos.\n",
        "\n",
        "- Esto se debe a que nuestro dataset tiene un importante desabalance entre las clases, y la que nos interesa es particulamente la clase minoritaria.\n",
        "\n",
        "- Preferimos trabajar con los datos tal como están en la base, para no introducir sesgos indeseados en los resultados. "
      ],
      "id": "9DhktGIjBKjr"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "W46aPETCYBLs"
      },
      "source": [
        "NUEVOS MODELOS:\n",
        "\n",
        "- Para mejorar el modelo baseline se realizaron distintas selecciones/incorporaciones de variables, y las modificaciones que surgieron como pertinentes según cada caso puntual. \n",
        "\n",
        "- Se detallan a continuación los modelos, análisis y conclusiones obtenidas."
      ],
      "id": "W46aPETCYBLs"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "G0TdNxZ3Z716"
      },
      "source": [
        "## Modelo 0: Regresión Logística \"Baseline retocado\"\n",
        "\n",
        "* Incluye ejercicios: [1,3,9,10,11,20] --> considerados difíciles. Se agregó el ejercicio 1 porque fue salteado por un gran porcentaje de estudiantes.\n",
        "* Imputación de NaN's en las variables tiempo invertido, e intentos fallidos (para cada ejercicio incluido) con: 0\n",
        "* Sin variable categórica (hizo/no hizo cada ejercicio incluido)"
      ],
      "id": "G0TdNxZ3Z716"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "QY8img8UaZX9",
        "cellView": "form"
      },
      "source": [
        "#@title\n",
        "# Regresión logística para ejercicios seleccionados (como difíciles), sin columna de último estado:\n",
        "X_train_baseline, X_test_baseline, y_train_baseline, y_test_baseline, y_train_pred_lr_baseline, y_test_pred_lr_baseline = construir_regresion_logistica([1,3,9,10,11,20], 0.3, True, 0, True)"
      ],
      "id": "QY8img8UaZX9",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "FBD4vcZobUTg"
      },
      "source": [
        "*Notar: El F1 score que viene en sklearn es la métrica de la clase 1.*"
      ],
      "id": "FBD4vcZobUTg"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "2PSFydTsbUTh",
        "cellView": "form",
        "outputId": "4922d977-b2d4-41cd-86e3-58ef83dfe05a"
      },
      "source": [
        "#@title\n",
        "print(\"Métricas del TRAIN para BASELINE\")\n",
        "print(classification_report(y_train_baseline, y_train_pred_lr_baseline))"
      ],
      "id": "2PSFydTsbUTh",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Métricas del TRAIN para BASELINE\n",
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.81      0.92      0.86       146\n",
            "           1       0.56      0.32      0.41        47\n",
            "\n",
            "    accuracy                           0.77       193\n",
            "   macro avg       0.68      0.62      0.63       193\n",
            "weighted avg       0.75      0.77      0.75       193\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "LEZZCTHGbUTh",
        "cellView": "form",
        "outputId": "014c413a-dd7e-49f2-97be-d4638c137dbb"
      },
      "source": [
        "#@title\n",
        "print(\"Métricas del TEST para BASELINE\")\n",
        "print(classification_report(y_test_baseline, y_test_pred_lr_baseline))"
      ],
      "id": "LEZZCTHGbUTh",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Métricas del TEST para BASELINE\n",
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.86      0.97      0.91        65\n",
            "           1       0.82      0.47      0.60        19\n",
            "\n",
            "    accuracy                           0.86        84\n",
            "   macro avg       0.84      0.72      0.76        84\n",
            "weighted avg       0.85      0.86      0.84        84\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "xpo4ts_Fffu4"
      },
      "source": [
        "### **Observaciones Modelo Baseline**\n",
        "\n",
        "* Las métricas del conjunto de Test, es decir, las medidas de performance del modelo cuando utilizamos datos nuevos no empleados en el entrenamiento, nos permiten decir lo siguiente:\n",
        "\n",
        "* Recall 0.47 en clase 1 --> significa que, si fueran 100 los desertores reales, lograríamos predecir correctamente a 47 (nos quedarían 53 desertores reales fuera de la predicción). No nos parece un buen resultado.\n",
        "\n",
        "* Precision 0.82 en clase 1 --> significa que, de 100 estudiantes predichos como desertores, 82 serían efectivamente desertores. Nos parece un buen resultado.\n",
        "\n",
        "* **Creemos que para nuestro objetivo general, es más importante el recall, ya que resulta más grave que el modelo NO prediga 53 de 100 desertores reales (porque no podríamos retenerlos); a que prediga erróneamente a 18 estudiantes como desertores (y que los mismos continúen en la plataforma).**\n",
        "\n",
        "* Obtenemos buenas métricas en la clase 0 (no desertores).\n",
        "\n",
        "* Obtenemos un Accuracy de 0.86 que consideramos bueno (significa que el modelo predice correctamente el 86% de los casos de ambas clases -es decir, el porcentaje considera en conjunto: casos de la clase 0 predichos como 0, y de la clase 1 predichos como 1-).\n",
        "\n",
        "* En resumen, si bien nos interesan medidas globales como Accuracy, y el F1 score “global” combinado para ambas clases (que considera Precision y Recall al mismo tiempo); para nuestro objetivo general nos parece MUY IMPORTANTE mejorar puntualmente la métrica RECALL para la CLASE 1.\n",
        "\n",
        "* **Esto es así, porque tenemos un interés \"pedagógico” en descubrir los desertores de la plataforma MUMUKI antes que abandonen; lo que permitiría implementar alguna estrategia de retención de forma oportuna.**\n"
      ],
      "id": "xpo4ts_Fffu4"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "7E-GsFkWnoP2"
      },
      "source": [
        "## Modelo 1: Regresión Logística \"Primeros 20\"\n",
        "\n",
        "* Incluye ejercicios: 1 al 20\n",
        "    \n",
        "* Imputación de NaN's en las variables tiempo invertido, e intentos fallidos (para cada ejercicio incluido) con: 0\n",
        "* Sin variable categórica (hizo/no hizo cada ejercicio incluido)\n"
      ],
      "id": "7E-GsFkWnoP2"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "BJSFcdlAr5HG",
        "cellView": "form"
      },
      "source": [
        "#@title\n",
        "# Regresión logística\n",
        "X_train_1_20, X_test_1_20, y_train_1_20, y_test_1_20, y_train_pred_lr_1_20, y_test_pred_lr_1_20 = construir_regresion_logistica(list(range(1,21)), 0.3, True, 0, True)"
      ],
      "id": "BJSFcdlAr5HG",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "zA5RjT86r5HH",
        "cellView": "form",
        "outputId": "75813a87-fa48-465c-df83-d3f69f901ca1"
      },
      "source": [
        "#@title\n",
        "print(\"Métricas del TRAIN para primeros 20 ejercicios, SIN columna 'ultimo_estado'\")\n",
        "print(classification_report(y_train_1_20, y_train_pred_lr_1_20))"
      ],
      "id": "zA5RjT86r5HH",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Métricas del TRAIN para primeros 20 ejercicios, SIN columna 'ultimo_estado'\n",
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.95      0.95      0.95       149\n",
            "           1       0.85      0.85      0.85        46\n",
            "\n",
            "    accuracy                           0.93       195\n",
            "   macro avg       0.90      0.90      0.90       195\n",
            "weighted avg       0.93      0.93      0.93       195\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "6WkTFqktr5HH",
        "cellView": "form",
        "outputId": "37b44353-05a2-437f-dace-2bceff7d0c57"
      },
      "source": [
        "#@title\n",
        "print(\"Métricas del TEST para primeros 20 ejercicios, SIN columna 'ultimo_estado'\")\n",
        "print(classification_report(y_test_1_20, y_test_pred_lr_1_20))"
      ],
      "id": "6WkTFqktr5HH",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Métricas del TEST para primeros 20 ejercicios, SIN columna 'ultimo_estado'\n",
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.87      0.89      0.88        62\n",
            "           1       0.67      0.64      0.65        22\n",
            "\n",
            "    accuracy                           0.82        84\n",
            "   macro avg       0.77      0.76      0.77        84\n",
            "weighted avg       0.82      0.82      0.82        84\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "p6WpNzsP05b4"
      },
      "source": [
        "### Comparación: baseline con modelo 1\n",
        "\n",
        "Vemos que el modelo 1 mejora el F1 score de la clase 1 en el conjunto de test (de 0.60 del baseline a 0.65 en modelo 1), debido a una mejora del RECALL de la clase 1 (lo que deseábamos, pasa de 0.47 a 0.64). Sin embargo, al mismo tiempo cae la precision de dicha clase, de 0.82 a 0.67 (consecuencia no deseada).\n",
        "\n"
      ],
      "id": "p6WpNzsP05b4"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wMkOjW7ao8zn"
      },
      "source": [
        "## Modelo 2: Regresión Logística \"Primeros 15\"\n",
        "\n",
        "* Incluye ejercicios: 1 al 15 \n",
        "* Sin variable categórica (hizo/no hizo cada ejercicio incluido)\n",
        "* Imputación de NaN's en las variables tiempo invertido, e intentos fallidos, para cada ejercicio incluido) con: 0"
      ],
      "id": "wMkOjW7ao8zn"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "R3zv2xvRo8zn",
        "cellView": "form"
      },
      "source": [
        "#@title\n",
        "# Regresión logística para los primeros 15 ejercicios, sin columna de último estado:\n",
        "X_train_1_15, X_test_1_15, y_train_1_15, y_test_1_15, y_train_pred_lr_1_15, y_test_pred_lr_1_15 = construir_regresion_logistica(list(range(1,16)), 0.3, True, 0, True)"
      ],
      "id": "R3zv2xvRo8zn",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "JwNmXjVMo8zn",
        "cellView": "form",
        "outputId": "573bf7f8-ece0-4868-8e4e-59845a234ca8"
      },
      "source": [
        "#@title\n",
        "print(\"Métricas del TRAIN para primeros 15 ejercicios, SIN columna 'ultimo_estado'\")\n",
        "print(classification_report(y_train_1_15, y_train_pred_lr_1_15))"
      ],
      "id": "JwNmXjVMo8zn",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Métricas del TRAIN para primeros 15 ejercicios, SIN columna 'ultimo_estado'\n",
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.89      0.93      0.91       149\n",
            "           1       0.74      0.61      0.67        46\n",
            "\n",
            "    accuracy                           0.86       195\n",
            "   macro avg       0.81      0.77      0.79       195\n",
            "weighted avg       0.85      0.86      0.85       195\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "ZvOqL7oCo8zo",
        "cellView": "form",
        "outputId": "7ca5b73c-a68b-461f-c6a5-d30a7b5cbec2"
      },
      "source": [
        "#@title\n",
        "print(\"Métricas del TEST para primeros 15 ejercicios, SIN columna 'ultimo_estado'\")\n",
        "print(classification_report(y_test_1_15, y_test_pred_lr_1_15))"
      ],
      "id": "ZvOqL7oCo8zo",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Métricas del TEST para primeros 15 ejercicios, SIN columna 'ultimo_estado'\n",
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.85      0.85      0.85        62\n",
            "           1       0.59      0.59      0.59        22\n",
            "\n",
            "    accuracy                           0.79        84\n",
            "   macro avg       0.72      0.72      0.72        84\n",
            "weighted avg       0.79      0.79      0.79        84\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "SfdGqGadpbnp"
      },
      "source": [
        "### Comparación: baseline, modelo 1 y modelo 2\n",
        "\n",
        "* En general, ver métricas para el conjunto de test:\n",
        "\n",
        "* Respecto al baseline, el modelo 2 muestra particularmente un mejor RECALL para la clase 1; pero caen el F1 de clase 0, 1, y por tanto su promedio, y también cae el accuracy). Es decir, en el sentido o \"interés específico deseado\" mejora, pero se pierde en los otros.\n",
        "\n",
        "* Luego, entre el modelo 1 y el 2, al bajar la cantidad de ejercicios incluidos, caen: accuracy, F1 global y F1 de la clase 1. También vemos que cae el RECALL de la clase 1, métrica más relacionada con nuestro problema principal. **Notar: también probamos con 11 ejercicios acumulados y las métricas fueron aún peores.\n",
        "\n",
        "* **En síntesis, el modelo 1 está mostrando la mejor performance hasta ahora.  Sin embargo, esto implica esperar la realización del ejercicio 20, en lugar de tener predicciones cuando los estudiantes lleguen, por ejemplo, al ejercicio 15. **\n",
        "\n",
        "* **Estamos observando un \"trade-off\" o intercambio entre una buena performance del modelo, y el tiempo de espera de resolución de ejercicios para tener las predicciones**. Es decir, incluir ejercicios con id mayores (más avanzados en el recorrido total), darían mejores métricas, pero implica esperar más tiempo para implementar medidas pedagógicas que eviten la desersión.\n",
        "\n",
        "* Notar: el recorrido total hasta ahora lo hemos \"medido\" en términos de ejericios (43 ejercicios únicos). Y estuvimos analizando contar con predicciones antes de la mitad del recorrido... Ahora nos interesa tomar también como referencia de \"recorrido\" el TIEMPO REAL EN DÍAS que implica la realización de 15 vs 20 ejercicios (u otra cantidad), para poder elegir el mejor modelo de predicción.\n",
        "\n",
        "* **Debido a este trade-off, nos preguntamos**:\n",
        "\n",
        "  - ¿cuánto tiempo promedio en días, implica esperar la resolución del ejercicio 20, respecto a la resolución del ejercicio 15, para obtener mejores predicciones de deserción? \n",
        "\n",
        "  - ¿Cuánto implica dicha espera en términos del tiempo promedio total de cursado?\n",
        "\n",
        "  - Mostramos a continuación, el análisis que responden a estas preguntas."
      ],
      "id": "SfdGqGadpbnp"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "BqsOI807vdF3"
      },
      "source": [
        "### Análisis: tiempo calendario\n",
        "\n",
        "**Cohortes en nuestra base de datos, y criterios de tiempo utilizadas para definirlas:**\n",
        "\n",
        "\n",
        "**III-2019:**\n",
        "\n",
        "* Fecha minima de envio: 2019-09-04 01:29:22.351000\n",
        "* Fecha maxima de envio: 2019-12-09 13:47:44.655000\n",
        "\n",
        "**II-2020:**\n",
        "\n",
        "* Fecha minima de envio: 2020-05-04 17:34:30.088000\n",
        "* Fecha maxima de envio: 2020-08-24 23:55:25.598000\n",
        "\n",
        "**III-2020:** \n",
        "\n",
        "* Fecha minima de envio: 2020-09-01 17:28:31.949000\n",
        "* Fecha maxima de envio: 2020-11-23 21:52:17.944000"
      ],
      "id": "BqsOI807vdF3"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "xzy2Tamez5D6",
        "cellView": "form"
      },
      "source": [
        "#@title\n",
        "# Nos quedamos con la fecha del primer envío de cada estudiante en cada ejercicio\n",
        "df_primeros_envios = df_submissions.sort_values([\"student_uid\", \"submission_created_at\"]).drop_duplicates([\"student_uid\", \"exercise_new_id\"])[[\"student_uid\",\"submission_created_at\", \"exercise_new_id\"]]\n",
        "\n",
        "# Pasamos los exercise_new_id a string para poder construir los nombres de las futuras columnas de la tabla pivot\n",
        "df_primeros_envios[\"exercise_new_id\"] = df_primeros_envios[\"exercise_new_id\"].astype(str)\n",
        "\n",
        "# Pivoteamos la información\n",
        "envios_ej_15_20 = pd.pivot(df_primeros_envios[df_primeros_envios.exercise_new_id.isin([\"15\", \"20\"])], index= [\"student_uid\"], columns = [\"exercise_new_id\"])\n",
        "\n",
        "\n",
        "envios_ej_15_20.columns = ['_'.join(col).strip() for col in envios_ej_15_20.columns.values]\n",
        "envios_ej_15_20.reset_index(inplace = True)\n",
        "\n",
        "# Agregamos una columna indicando el cuatrimestre (cohorte) al que pertenecía el estudiante\n",
        "envios_ej_15_20[\"mes_anio\"] = np.where((envios_ej_15_20.submission_created_at_15 < '2020-01-01')|(envios_ej_15_20.submission_created_at_20 < '2020-01-01' ), \"I\", np.where((envios_ej_15_20.submission_created_at_15 < '2020-08-30')|(envios_ej_15_20.submission_created_at_20 < '2020-08-30'), \"II\", \"III\"))\n",
        "\n",
        "# Agregamos la fecha de inicio y fin de cada cohorte\n",
        "envios_ej_15_20[\"inicio\"] = np.where(envios_ej_15_20.mes_anio == 'I', pd.to_datetime(\"2019,09,04\", format=\"%Y,%m,%d\"), np.where(envios_ej_15_20.mes_anio == 'II', pd.to_datetime(\"2020,05,04\", format=\"%Y,%m,%d\"), pd.to_datetime(\"2020,09,01\", format=\"%Y,%m,%d\")))\n",
        "envios_ej_15_20[\"fin\"] = np.where(envios_ej_15_20.mes_anio == 'I', pd.to_datetime(\"2019,12,09\", format=\"%Y,%m,%d\"), np.where(envios_ej_15_20.mes_anio == 'II', pd.to_datetime(\"2020,08,24\", format=\"%Y,%m,%d\"), pd.to_datetime(\"2020,11,23\", format=\"%Y,%m,%d\")))\n",
        "\n",
        "# Calculamos la cantidad de días entre el inicio y fin de cada curso\n",
        "envios_ej_15_20[\"duracion_curso\"] = (envios_ej_15_20.fin - envios_ej_15_20.inicio).dt.days\n",
        "\n",
        "# Calculamos la cantidad de días entre el inicio y la realización de cada ejercicio\n",
        "envios_ej_15_20[\"tiempo_inicio_al_15\"] = (envios_ej_15_20.submission_created_at_15 - envios_ej_15_20.inicio).dt.days\n",
        "envios_ej_15_20[\"tiempo_inicio_al_20\"] = (envios_ej_15_20.submission_created_at_20 - envios_ej_15_20.inicio).dt.days\n",
        "\n",
        "envios_ej_15_20"
      ],
      "id": "xzy2Tamez5D6",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 355
        },
        "id": "yrEltnFu2cYY",
        "cellView": "form",
        "collapsed": true,
        "outputId": "4690fa17-b1e1-41b5-da61-1db774da1150"
      },
      "source": [
        "#@title\n",
        "# En el siguiente gráfico se ve la distribución de la fecha de inicio del ejercicio 15 entre los estudiantes\n",
        "\n",
        "envios_ej_15_20.plot.scatter(y=\"inicio\", x=\"submission_created_at_15\", figsize=(12,5))"
      ],
      "id": "yrEltnFu2cYY",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7fc45f0ca2d0>"
            ]
          },
          "metadata": {},
          "execution_count": 50
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 864x360 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 355
        },
        "id": "ZQw17vCP-npL",
        "cellView": "form",
        "collapsed": true,
        "outputId": "46bc9d33-6d58-4a25-c84d-ceb6c1962432"
      },
      "source": [
        "#@title\n",
        "# En el siguiente gráfico se ve la distribución de la fecha de inicio del ejercicio 20 entre los estudiantes\n",
        "\n",
        "envios_ej_15_20.plot.scatter(y=\"inicio\", x=\"submission_created_at_20\", figsize=(12,5))"
      ],
      "id": "ZQw17vCP-npL",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7fc45eade510>"
            ]
          },
          "metadata": {},
          "execution_count": 51
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 864x360 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "Xi6aoikLE6x5",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 166
        },
        "cellView": "form",
        "collapsed": true,
        "outputId": "819350f0-3432-4208-de89-f4227e9f74a9"
      },
      "source": [
        "#@title\n",
        "envios_ej_15_20[envios_ej_15_20.mes_anio == \"II\"].sort_values(\"submission_created_at_15\", ascending = False).head(2)"
      ],
      "id": "Xi6aoikLE6x5",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>student_uid</th>\n",
              "      <th>submission_created_at_15</th>\n",
              "      <th>submission_created_at_20</th>\n",
              "      <th>mes_anio</th>\n",
              "      <th>inicio</th>\n",
              "      <th>fin</th>\n",
              "      <th>duracion_curso</th>\n",
              "      <th>tiempo_inicio_al_15</th>\n",
              "      <th>tiempo_inicio_al_20</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>140</th>\n",
              "      <td>253</td>\n",
              "      <td>2020-08-14 23:35:47.077</td>\n",
              "      <td>2020-08-18 21:27:28.657</td>\n",
              "      <td>II</td>\n",
              "      <td>2020-05-04</td>\n",
              "      <td>2020-08-24</td>\n",
              "      <td>112</td>\n",
              "      <td>102.0</td>\n",
              "      <td>106.0</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>139</th>\n",
              "      <td>252</td>\n",
              "      <td>2020-08-09 06:09:54.837</td>\n",
              "      <td>2020-08-09 06:24:11.510</td>\n",
              "      <td>II</td>\n",
              "      <td>2020-05-04</td>\n",
              "      <td>2020-08-24</td>\n",
              "      <td>112</td>\n",
              "      <td>97.0</td>\n",
              "      <td>97.0</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "     student_uid  ... tiempo_inicio_al_20\n",
              "140          253  ...               106.0\n",
              "139          252  ...                97.0\n",
              "\n",
              "[2 rows x 9 columns]"
            ]
          },
          "metadata": {},
          "execution_count": 52
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "nx83teYqyzbw",
        "cellView": "form"
      },
      "source": [
        "#@title\n",
        "envios_ej_15_20 = envios_ej_15_20[~envios_ej_15_20.student_uid.isin([252, 253])]"
      ],
      "id": "nx83teYqyzbw",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "odCohcOXBHvL",
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 300
        },
        "cellView": "form",
        "outputId": "3609725e-3e7d-4bfa-9d56-5ee01a5aabab"
      },
      "source": [
        "#@title\n",
        "# Descriptivas de las variables (para todos los estudiantes): \n",
        "\n",
        "# * duración total del curso (inicio = primer envío; fin = último envío -dentro del cuatrimestre calendario correspondiente-)\n",
        "\n",
        "# * tiempo transcurrido entre el inicio del curso y la resolución del ejercicio 15\n",
        "\n",
        "# * tiempo transcurrido entre el inicio del curso y la resolución del ejercicio 20\n",
        "\n",
        "envios_ej_15_20.describe()"
      ],
      "id": "odCohcOXBHvL",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/html": [
              "<div>\n",
              "<style scoped>\n",
              "    .dataframe tbody tr th:only-of-type {\n",
              "        vertical-align: middle;\n",
              "    }\n",
              "\n",
              "    .dataframe tbody tr th {\n",
              "        vertical-align: top;\n",
              "    }\n",
              "\n",
              "    .dataframe thead th {\n",
              "        text-align: right;\n",
              "    }\n",
              "</style>\n",
              "<table border=\"1\" class=\"dataframe\">\n",
              "  <thead>\n",
              "    <tr style=\"text-align: right;\">\n",
              "      <th></th>\n",
              "      <th>student_uid</th>\n",
              "      <th>duracion_curso</th>\n",
              "      <th>tiempo_inicio_al_15</th>\n",
              "      <th>tiempo_inicio_al_20</th>\n",
              "    </tr>\n",
              "  </thead>\n",
              "  <tbody>\n",
              "    <tr>\n",
              "      <th>count</th>\n",
              "      <td>229.000000</td>\n",
              "      <td>229.000000</td>\n",
              "      <td>224.000000</td>\n",
              "      <td>215.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>mean</th>\n",
              "      <td>219.698690</td>\n",
              "      <td>97.388646</td>\n",
              "      <td>21.191964</td>\n",
              "      <td>23.730233</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>std</th>\n",
              "      <td>76.998069</td>\n",
              "      <td>13.007489</td>\n",
              "      <td>15.526160</td>\n",
              "      <td>16.822097</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>min</th>\n",
              "      <td>76.000000</td>\n",
              "      <td>83.000000</td>\n",
              "      <td>0.000000</td>\n",
              "      <td>0.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>25%</th>\n",
              "      <td>158.000000</td>\n",
              "      <td>83.000000</td>\n",
              "      <td>11.000000</td>\n",
              "      <td>11.500000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>50%</th>\n",
              "      <td>219.000000</td>\n",
              "      <td>96.000000</td>\n",
              "      <td>19.000000</td>\n",
              "      <td>20.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>75%</th>\n",
              "      <td>286.000000</td>\n",
              "      <td>112.000000</td>\n",
              "      <td>22.000000</td>\n",
              "      <td>28.000000</td>\n",
              "    </tr>\n",
              "    <tr>\n",
              "      <th>max</th>\n",
              "      <td>353.000000</td>\n",
              "      <td>112.000000</td>\n",
              "      <td>83.000000</td>\n",
              "      <td>83.000000</td>\n",
              "    </tr>\n",
              "  </tbody>\n",
              "</table>\n",
              "</div>"
            ],
            "text/plain": [
              "       student_uid  duracion_curso  tiempo_inicio_al_15  tiempo_inicio_al_20\n",
              "count   229.000000      229.000000           224.000000           215.000000\n",
              "mean    219.698690       97.388646            21.191964            23.730233\n",
              "std      76.998069       13.007489            15.526160            16.822097\n",
              "min      76.000000       83.000000             0.000000             0.000000\n",
              "25%     158.000000       83.000000            11.000000            11.500000\n",
              "50%     219.000000       96.000000            19.000000            20.000000\n",
              "75%     286.000000      112.000000            22.000000            28.000000\n",
              "max     353.000000      112.000000            83.000000            83.000000"
            ]
          },
          "metadata": {},
          "execution_count": 54
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "gg8u0u2T0zdB"
      },
      "source": [
        "**Síntesis de las descriptivas para todos los estudiantes analizados en conjunto**\n",
        "\n",
        "* En promedio, habría que esperar **21 días** desde el comienzo de los envíos, para tener información de las resoluciones del ejercicio 15. Y **24 días** para tener las del ejercicio 20. Es decir, habría que considerar un promedio de **3 días de diferencia a esperar** entre dichos ejercicios.\n",
        "\n",
        "* Como se puede apreciar una **alta variabilidad** en las distribuciones de estos tiempos (16 y 17 días de desviación alrededor de la medias referenciadas), observamos también las medianas y el cuartil 3 de cada distribución:  \n",
        "  \n",
        "  - **el valor central** de la distribución de \"tiempo para llegar al ejercicio 15\" es de **19 días**; y \"para llegar al ejercicio 20\", de **20 días**.\n",
        "\n",
        "  - **el tercer cuartil** es **22 días** entre el incio y el ejercicio 15, y de **28 días** entre el inicio y el ejercicio 20.\n",
        "\n",
        "  - Es decir, **entre las resoluciones del ejercicio 15 y 20 hay un día de diferencia, pero se ve una mayor dispersión de datos** respecto al tercer cuartil cuando se mira la **resolución del ejercicio 20** (3 días de desviación entre Q2 y Q3 para llegar al ejercicio 15, y 8 días de desviación entre Q2 y Q3 para llegar al ejercicio 15).\n",
        "\n",
        "* En cualquier caso, observamos que la **mediana** en la **duración total** del curso es de **96 días**, y el máximo tiempo alcanzado para la finalización fueron 112 días.\n",
        "\n",
        "* **En síntesis, a pesar de haber cierta variabilidad en la distribución del tiempo de resolución entre los estudiantes, podemos ver que no resulta tan significativo esperar hasta 3 días en promedio para contar con la resolución del ejercicio 20. O quizá hasta un máximo de 8 días para contar con el 75% de las resoluciones de dicho ejercicio, considerando que el recorrido total es alcanzado por el 75% de los estudiantes en 112 días.**\n",
        "\n",
        "* Datos:\n",
        "  - Para contar con la resolución del ejercicio 15 --> 22 días/112 días = 20% del tiempo total.\n",
        "  - Para contar con la resolución del ejercicio 20 --> 28 días/112 días = 25% del tiempo total.\n",
        "  - Diferencia de días máxima entre el 15 y el 20 --> 8/112 = 7%\n",
        "\n",
        "* Con todo esto consideramos que, esperar entre 1 y máximo 8 días, puede ser aceptable para contar con mayor cantidad de ejercicios (y por ello mejores predicciones). Ya que esta \"primera mitad de ejercicios\" (1 al 20) son resueltos por los estudiantes en el primer quinto del tiempo total del cursado.\n",
        "\n",
        "\n"
      ],
      "id": "gg8u0u2T0zdB"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 354
        },
        "id": "K9jTer2l-vEi",
        "cellView": "form",
        "collapsed": true,
        "outputId": "a07ea775-2048-488e-9b94-d9eef44b0014"
      },
      "source": [
        "#@title\n",
        "plt.figure(figsize = (10,5))\n",
        "sns.boxplot(data = envios_ej_15_20, x = \"tiempo_inicio_al_15\", y=\"mes_anio\")"
      ],
      "id": "K9jTer2l-vEi",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7f7210eb8850>"
            ]
          },
          "metadata": {},
          "execution_count": 26
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x360 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/",
          "height": 354
        },
        "id": "RE49N3NVAVSi",
        "collapsed": true,
        "cellView": "form",
        "outputId": "52b6e4ca-401a-4567-a8cb-a75e57086899"
      },
      "source": [
        "#@title\n",
        "\n",
        "plt.figure(figsize = (10,5))\n",
        "sns.boxplot(data = envios_ej_15_20, x = \"tiempo_inicio_al_20\", y=\"mes_anio\")"
      ],
      "id": "RE49N3NVAVSi",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "<matplotlib.axes._subplots.AxesSubplot at 0x7f7210d6f850>"
            ]
          },
          "metadata": {},
          "execution_count": 27
        },
        {
          "output_type": "display_data",
          "data": {
            "image/png": "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\n",
            "text/plain": [
              "<Figure size 720x360 with 1 Axes>"
            ]
          },
          "metadata": {
            "needs_background": "light"
          }
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "0lrT3Tff1jkc"
      },
      "source": [
        "**Observaciones de las descriptivas discriminadas por cohortes de estudiantes**\n",
        "\n",
        "* Comparativamente, vemos que el tiempo \"mediano\" en el que se comienzan a resolver cada uno de los ejercicios disminuye entre las sucesivas cohortes. \n",
        "\n",
        "* La cohorte I tiene aprox. la mitad de estudiantes que las cohortes II y III (46, 93 y 90 estudiantes respectivamente). Las últimas cohortes podrían considerarse más representativas en términos de cantidad de casos.\n",
        "\n",
        "* Así como se observó en las descriptivas generales, en estas descriptivas por cohortes también se aprecia una mayor dispersión en los tiempos que tardan los estudiantes en llegar al ejercicio 20, vs. llegar al ejercicio 15.\n",
        "\n",
        "* A pesar de las diferencias y variabilidades resaltadas, arribamos a las mismas conclusiones obtenidas con las descriptivas generales, en relación a nuestro problema de estudio y a las preguntas que nos habíamos planteado analizar."
      ],
      "id": "0lrT3Tff1jkc"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "peE18Y3sd_5A"
      },
      "source": [
        "\n",
        "**Observaciones generales del análisis del tiempo calendario**\n",
        "\n",
        " 1) Haber descubierto que esperar entre uno a ocho días para obtener mejores predicciones resulta relativamente insignificante con respecto a la \"duración total\" del curso (112 días), es una muy buena información a considerar. \n",
        "\n",
        " 2) De todas formas, nos interesa seguir probando versiones del modelo, que permitan disminuir la cantidad de ejercicios necesarios para obtener predicciones, ya que \"esperar\" hasta el ejercicio 20 igual implica contar con aproximadamente \"la mitad del recorrido total en términos de actividades\".\n"
      ],
      "id": "peE18Y3sd_5A"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EFdE7XhgoDX-"
      },
      "source": [
        "### Análisis: curación de variables utilizadas"
      ],
      "id": "EFdE7XhgoDX-"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "3g7gAUqUpVq7"
      },
      "source": [
        "**Observamos al realizar las estimaciones, que en los 2 modelos anteriores no es muy adecuado completar con 0's las celdas vacías** de cada una de las variables de intentos incorrectos y de tiempo de resolución. Dichos valores perdidos, corresponden a **estudiantes que no resolvieron esos ejercicios**, y al completarlos con 0's, los estaríamos **confundiendo con los que sí los resolvieron correctamente en el primer envío**.\n",
        "\n",
        "Por lo tanto, realizaremos un nuevo análisis donde no completaremos los `NaN`'s con `0`, si no que los dejaremos como están para decidir qué otro criterio aplicar (seleccionar/eliminar variables, o conservar las variables y decidir respecto a su imputación).\n",
        "\n",
        "**IMPORTANTE:**\n",
        "\n",
        "* **Necesitamos imputar los valores perdidos de las variables de intentos fallidos y de tiempo invertido, para cada ejercicio, que decidamos mantener en el dataset** (es decir, aquellas que tienen aproximadamente menos del 20% de datos perdidos).\n",
        "\n",
        "* Para poder decidir un criterio de imputación, necesitamos hacer un supuesto respecto a qué representan los casos perdidos, y tomar una decisión.\n",
        "\n",
        "* Describimos a continuación la línea de pensamiento que hicimos y su conclusión:\n",
        "\n",
        "  1) Los casos perdidos de las variables intentos fallidos y tiempo invertido, son aquellos estudiantes que no enviaron resoluciones para determinados ejercicios.\n",
        "  \n",
        "  2) Si no enviaron soluciones, es más probable que sea porque no los sabían hacer (ejercicios difíciles para ellos), a que sea porque sí los sabían hacer.\n",
        "\n",
        "  3) Luego, los estudiantes que saltearon ejercicios por ser “difíciles”, serían más parecidos a un perfil desertor, que a un perfil no desertor. \n",
        "\n",
        "  4) Por lo visto en entregas anteriores, el perfil desertor se condice con valores altos de intentos fallidos y tiempo invertido.\n",
        "\n",
        "  **5) Por lo tanto, sería lógico imputar los `NaN`'s con valores similares al del perfil desertor.**\n",
        "  \n",
        "  6) En este sentido, consideramos que para tratar de mantener la forma de la distribución de cada variable, y no sesgar con la imputación el modelo predictivo, **deberíamos asignar los`NaN`'s con el valor máximo de cada variable.**"
      ],
      "id": "3g7gAUqUpVq7"
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "id": "Sm8e6J4IpBVz",
        "outputId": "17c78563-80ef-429a-da43-b6c867ac9035"
      },
      "source": [
        "data_por_estudiante = data_por_student(list(range(1, 21)), np.nan, np.nan)\n",
        "\n",
        "# Analizamos porcentaje de datos perdidos por cada variable\n",
        "(data_por_estudiante.isna().sum() / 277).sort_values(ascending = False)"
      ],
      "id": "Sm8e6J4IpBVz",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "execute_result",
          "data": {
            "text/plain": [
              "intentos_incorrectos_18     0.350181\n",
              "tiempo_total_18             0.350181\n",
              "tiempo_total_8              0.277978\n",
              "intentos_incorrectos_8      0.277978\n",
              "tiempo_total_1              0.256318\n",
              "intentos_incorrectos_1      0.256318\n",
              "tiempo_total_20             0.223827\n",
              "intentos_incorrectos_20     0.223827\n",
              "intentos_incorrectos_19     0.202166\n",
              "tiempo_total_16             0.202166\n",
              "intentos_incorrectos_16     0.202166\n",
              "tiempo_total_19             0.202166\n",
              "tiempo_total_15             0.191336\n",
              "intentos_incorrectos_15     0.191336\n",
              "intentos_incorrectos_13     0.180505\n",
              "tiempo_total_13             0.180505\n",
              "tiempo_total_14             0.173285\n",
              "intentos_incorrectos_14     0.173285\n",
              "intentos_incorrectos_12     0.151625\n",
              "tiempo_total_12             0.151625\n",
              "tiempo_total_11             0.140794\n",
              "intentos_incorrectos_11     0.140794\n",
              "tiempo_total_10             0.111913\n",
              "intentos_incorrectos_10     0.111913\n",
              "tiempo_total_6              0.083032\n",
              "intentos_incorrectos_6      0.083032\n",
              "tiempo_total_9              0.068592\n",
              "intentos_incorrectos_9      0.068592\n",
              "tiempo_total_3              0.028881\n",
              "intentos_incorrectos_3      0.028881\n",
              "tiempo_total_2              0.010830\n",
              "intentos_incorrectos_2      0.010830\n",
              "cond_cant_ejer_resueltos    0.000000\n",
              "ultimo_estado_9             0.000000\n",
              "ultimo_estado_8             0.000000\n",
              "ultimo_estado_1             0.000000\n",
              "ultimo_estado_10            0.000000\n",
              "ultimo_estado_11            0.000000\n",
              "ultimo_estado_12            0.000000\n",
              "ultimo_estado_13            0.000000\n",
              "ultimo_estado_14            0.000000\n",
              "ultimo_estado_15            0.000000\n",
              "ultimo_estado_16            0.000000\n",
              "ultimo_estado_18            0.000000\n",
              "ultimo_estado_19            0.000000\n",
              "ultimo_estado_2             0.000000\n",
              "ultimo_estado_20            0.000000\n",
              "ultimo_estado_3             0.000000\n",
              "ultimo_estado_6             0.000000\n",
              "student_uid                 0.000000\n",
              "dtype: float64"
            ]
          },
          "metadata": {},
          "execution_count": 25
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "_tBBlLQ95DfX"
      },
      "source": [
        "Si vemos por columna la cantidad de `NaN`'s, la mayoría se da en los ejercicios 18 (35%), 8 (27%), 1 (25%) y 20 (22%). \n",
        "\n",
        "Estos datos se corresponden con el porcentaje de estudiantes que no realizaron envíos en cada uno de estos ejercicios.\n",
        "\n",
        "**Dado que son porcentajes altos y no queremos sesgar los resultados del modelo, consideramos la posibilidad de eliminar estas variables.** Pero como tampoco queremos perder la información referida a los estudiantes que no hicieron envíos en estos ejercicios, entonces **pensamos en generar una variable dicotómica, que indique si hizo o no cada ejercicio**.\n",
        "\n",
        "Mostramos las métricas de dos modelos más, considerando estas opciones combinadas, para tomar una decisión final."
      ],
      "id": "_tBBlLQ95DfX"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "OWqIvpWk3Dn-"
      },
      "source": [
        "## Modelo 3: Regresión Logística \"Mejorada\" (post análisis) \n",
        "\n",
        "* Incluye ejercicios: **1 al 20 (quitando ejercicios 1, 8, 18, 20)**\n",
        "* Imputación de NaN's en las variables tiempo invertido, e intentos fallidos (para cada ejercicio incluido) con: **máximo valor de cada variable**\n",
        "* Con variable categórica (**hizo/no hizo cada ejercicio incluido**)"
      ],
      "id": "OWqIvpWk3Dn-"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "todyEatKIWkP",
        "cellView": "form"
      },
      "source": [
        "#@title\n",
        "# Generación del DF \n",
        "df_1_20_seleccionado = data_por_student(list(set(list(range(1,21)))), imputar_con_constante = False, imputacion_nan = 0)\n",
        "df_1_20_seleccionado = df_1_20_seleccionado.drop([\"intentos_incorrectos_1\", \"intentos_incorrectos_8\", \"intentos_incorrectos_18\", \"intentos_incorrectos_20\", \"tiempo_total_1\", \"tiempo_total_8\", \"tiempo_total_18\", \"tiempo_total_20\"], axis = 1)\n",
        "df_1_20_seleccionado"
      ],
      "id": "todyEatKIWkP",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "ErxBxLrdpUq8",
        "cellView": "form"
      },
      "source": [
        "#@title\n",
        "X_train_1_20_seleccionado, X_test_1_20_seleccionado, y_train_1_20_seleccionado, y_test_1_20_seleccionado, y_train_pred_lr_1_20_seleccionado, y_test_pred_lr_1_20_seleccionado = regresion_logistica(df_1_20_seleccionado, 0.3, False)"
      ],
      "id": "ErxBxLrdpUq8",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "MKuP7MEl3Dn_",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "cellView": "form",
        "outputId": "489f0d60-58e5-4386-a892-ab8f854dff96"
      },
      "source": [
        "#@title\n",
        "print(\"Métricas del TRAIN para 1 al 20 (quitando ejercicios 1, 8, 18, 20), CON columna 'ultimo_estado'\")\n",
        "print(classification_report(y_train_1_20_seleccionado, y_train_pred_lr_1_20_seleccionado))"
      ],
      "id": "MKuP7MEl3Dn_",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Métricas del TRAIN para 1 al 20 (quitando ejercicios 1, 8, 18, 20), CON columna 'ultimo_estado'\n",
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.97      0.99      0.98       149\n",
            "           1       0.98      0.91      0.94        46\n",
            "\n",
            "    accuracy                           0.97       195\n",
            "   macro avg       0.98      0.95      0.96       195\n",
            "weighted avg       0.97      0.97      0.97       195\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "f4iKAfLK3Dn_",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "cellView": "form",
        "outputId": "91b1106f-9915-40db-a511-8c1addbcf7d8"
      },
      "source": [
        "#@title\n",
        "print(\"Métricas del TEST para 1 al 20 (quitando ejercicios 1, 8, 18, 20), CON columna 'ultimo_estado'\")\n",
        "print(classification_report(y_test_1_20_seleccionado, y_test_pred_lr_1_20_seleccionado))"
      ],
      "id": "f4iKAfLK3Dn_",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Métricas del TEST para 1 al 20 (quitando ejercicios 1, 8, 18, 20), CON columna 'ultimo_estado'\n",
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.98      0.94      0.96        62\n",
            "           1       0.84      0.95      0.89        22\n",
            "\n",
            "    accuracy                           0.94        84\n",
            "   macro avg       0.91      0.95      0.93        84\n",
            "weighted avg       0.95      0.94      0.94        84\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "dXYt9ae4J4YZ"
      },
      "source": [
        "#### **Conclusión modelo 3**\n",
        "\n",
        "* Los valores de las métricas descriptas a continuación, las observamos en el conjunto de test (ya que allí vemos la performance del modelo ante nuevas observaciones)\n",
        "\n",
        "* Comparando con el modelo baseline, logramos una amplia mejora del RECALL de la clase 1 (de 0.47 a 0.95), aunque bajan se mantienen altas precision de clase 1 (de 0.84 a 0.82) y clase 0 (de 0.98 a 0.86). Mejora el recall de clase 0 de 0.94 a 0.97 y el accuracy en general (0.84 a 0.94).\n",
        "\n",
        "* **Con relación al modelo 1 (que hasta ahora era el mejor), se observan mejoras para todas las métricas (ver en particular la mejora del RECALL de la clase 1 => 0.64 a 0.95)**; y precision de clase 1 (0.67 a 0.84). Accuracy pasa de de 0.82 a 0.94, y las métricas de la clase 0 también mejoran significativamente en el modelo 3 (de valores cercanos a 0.88, se pasa a valores por encima del 0.94).\n",
        "\n",
        "* Si bien entre los datos del conjunto de train y test del Modelo 3, la precisión de la clase 1 baja, (de 0.98 a 0.84) se mantiene en un valor alto y es un resultado esperable, reflejando que el modelo no overfitea."
      ],
      "id": "dXYt9ae4J4YZ"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "EhsYtHNXIS9_"
      },
      "source": [
        "## Modelo 4: Regresión Logística (chequeo)\n",
        "\n",
        "\n",
        "* Incluye ejercicios: 1 al 20 (dejamos ejercicios 1, 8, 18, 20 con alto % Nan's)\n",
        "* Con variable categórica (hizo/no hizo cada ejercicio incluido)\n",
        "* Imputación de NaN's en las variables tiempo invertido, e intentos fallidos, para cada ejercicio incluido) con: máximo valor de cada variable "
      ],
      "id": "EhsYtHNXIS9_"
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "DaOdeX0UITRh",
        "cellView": "form"
      },
      "source": [
        "#@title\n",
        "# Regresión logística para los primeros 20 ejercicios, CON columna de último estado:\n",
        "X_train_1_20, X_test_1_20, y_train_1_20, y_test_1_20, y_train_pred_lr_1_20, y_test_pred_lr_1_20 = construir_regresion_logistica(list(range(1,21)), 0.3, False, 0, False)"
      ],
      "id": "DaOdeX0UITRh",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "4vjS8QGuITRj",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "cellView": "form",
        "outputId": "680aa8df-5081-4dba-c129-5a61eb0b2ad3"
      },
      "source": [
        "#@title\n",
        "print(\"Métricas del TRAIN para primeros 20 ejercicios, CON columna 'ultimo_estado'\")\n",
        "print(classification_report(y_train_1_20, y_train_pred_lr_1_20))"
      ],
      "id": "4vjS8QGuITRj",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Métricas del TRAIN para primeros 20 ejercicios, CON columna 'ultimo_estado'\n",
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.97      0.99      0.98       149\n",
            "           1       0.95      0.91      0.93        46\n",
            "\n",
            "    accuracy                           0.97       195\n",
            "   macro avg       0.96      0.95      0.96       195\n",
            "weighted avg       0.97      0.97      0.97       195\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "id": "mPcFN6QbITRk",
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "cellView": "form",
        "outputId": "66ef7c81-45b5-4fc8-d45c-e9aa79498a87"
      },
      "source": [
        "#@title\n",
        "print(\"Métricas del TEST para primeros 20 ejercicios, CON columna 'ultimo_estado'\")\n",
        "print(classification_report(y_test_1_20, y_test_pred_lr_1_20))"
      ],
      "id": "mPcFN6QbITRk",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Métricas del TEST para primeros 20 ejercicios, CON columna 'ultimo_estado'\n",
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       1.00      0.95      0.98        62\n",
            "           1       0.88      1.00      0.94        22\n",
            "\n",
            "    accuracy                           0.96        84\n",
            "   macro avg       0.94      0.98      0.96        84\n",
            "weighted avg       0.97      0.96      0.96        84\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "4oCN_hS7LEOD"
      },
      "source": [
        "#### **Conclusión modelo 4 (final)**\n",
        "\n",
        "**Vemos que es correcto eliminar los ejercicios que tuvieron alto porcentaje de NaN's, ya que dejándolos el modelo muestra problemas de overfitting en test.**\n",
        "\n",
        "Es decir, entendemos al ver estos resultados, que si imputamos las variables con muchos Nan con el perfil \"desertor\", en realidad lo que estamos haciendo es clasificar \"previamente\" a los estudiantes que no resolvieron ejercicios, como desertores. Y como son muchos estudiantes en esas variables, la regresión aprende el comportamiento \"imputado\" y luego arroja una \"métrica = 1\" en recall de clase 1 (también en precision de clase 0 resulta =1 y el recall de clase 0 es muy próximo a 1 = 0.95).\n",
        "\n",
        "**Por ende, el mejor modelo obtenido es el número 3**."
      ],
      "id": "4oCN_hS7LEOD"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "5SZmThBxGT53"
      },
      "source": [
        "## Modelo 3 (simplificado)\n",
        "Pensamos entonces en la posibilidad de quitar las variables categóricas de cada ejercicio (hizo/ no hizo) en el modelo 3, para ver si cambiaba la performance del mismo. \n",
        "\n",
        "* Incluye ejercicios: **1 al 20 (quitando ejercicios 1, 8, 18, 20)**\n",
        "* Imputación de NaN's en las variables tiempo invertido, e intentos fallidos (para cada ejercicio incluido) con: **máximo valor de cada variable**\n",
        "* Sin variable categórica (**hizo/no hizo cada ejercicio incluido**)"
      ],
      "id": "5SZmThBxGT53"
    },
    {
      "cell_type": "code",
      "metadata": {
        "cellView": "form",
        "id": "_YIucOibHJxL"
      },
      "source": [
        "#@title\n",
        "# Generación del DF \n",
        "df_1_20_seleccionado = data_por_student(list(set(list(range(1,21)))), imputar_con_constante = False, imputacion_nan = 0)\n",
        "df_1_20_seleccionado = df_1_20_seleccionado.drop([\"intentos_incorrectos_1\", \"intentos_incorrectos_8\", \"intentos_incorrectos_18\", \"intentos_incorrectos_20\", \"tiempo_total_1\", \"tiempo_total_8\", \"tiempo_total_18\", \"tiempo_total_20\"], axis = 1)\n",
        "df_1_20_seleccionado"
      ],
      "id": "_YIucOibHJxL",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "cellView": "form",
        "id": "_U1z250gHJxN"
      },
      "source": [
        "#@title\n",
        "X_train_1_20_seleccionado, X_test_1_20_seleccionado, y_train_1_20_seleccionado, y_test_1_20_seleccionado, y_train_pred_lr_1_20_seleccionado, y_test_pred_lr_1_20_seleccionado = regresion_logistica(df_1_20_seleccionado, 0.3, True)"
      ],
      "id": "_U1z250gHJxN",
      "execution_count": null,
      "outputs": []
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "cellView": "form",
        "id": "aEzTYZ0gHJxN",
        "outputId": "550b04ee-eeef-43c4-f7b6-feffe0fb6b4c"
      },
      "source": [
        "#@title\n",
        "print(\"Métricas del TRAIN para 1 al 20 (quitando ejercicios 1, 8, 18, 20), SIN columna 'ultimo_estado'\")\n",
        "print(classification_report(y_train_1_20_seleccionado, y_train_pred_lr_1_20_seleccionado))"
      ],
      "id": "aEzTYZ0gHJxN",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Métricas del TRAIN para 1 al 20 (quitando ejercicios 1, 8, 18, 20), SIN columna 'ultimo_estado'\n",
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.97      0.99      0.98       149\n",
            "           1       0.95      0.91      0.93        46\n",
            "\n",
            "    accuracy                           0.97       195\n",
            "   macro avg       0.96      0.95      0.96       195\n",
            "weighted avg       0.97      0.97      0.97       195\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "code",
      "metadata": {
        "colab": {
          "base_uri": "https://localhost:8080/"
        },
        "cellView": "form",
        "id": "kqKDyeOCHJxO",
        "outputId": "80825bb1-15c0-4bf0-b2b4-05e1040eaa73"
      },
      "source": [
        "#@title\n",
        "print(\"Métricas del TEST para 1 al 20 (quitando ejercicios 1, 8, 18, 20), SIN columna 'ultimo_estado'\")\n",
        "print(classification_report(y_test_1_20_seleccionado, y_test_pred_lr_1_20_seleccionado))"
      ],
      "id": "kqKDyeOCHJxO",
      "execution_count": null,
      "outputs": [
        {
          "output_type": "stream",
          "name": "stdout",
          "text": [
            "Métricas del TEST para 1 al 20 (quitando ejercicios 1, 8, 18, 20), SIN columna 'ultimo_estado'\n",
            "              precision    recall  f1-score   support\n",
            "\n",
            "           0       0.98      0.95      0.97        62\n",
            "           1       0.88      0.95      0.91        22\n",
            "\n",
            "    accuracy                           0.95        84\n",
            "   macro avg       0.93      0.95      0.94        84\n",
            "weighted avg       0.95      0.95      0.95        84\n",
            "\n"
          ]
        }
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "bvP6X5sYHDsH"
      },
      "source": [
        "**Obtuvimos métricas muy similares a las que mostramos en el modelo 3 original. Por lo tanto, priorizando el principio de parsimonia, entendemos que el mejor modelo de predicción de deserción es el \"Modelo 3 (simplificado)\".**"
      ],
      "id": "bvP6X5sYHDsH"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "t5ptlFcEFFyo"
      },
      "source": [
        "# 6- ANÁLISIS FUTUROS\n",
        "\n",
        "Se proponen como mejoras al modelo seleccionado los siguientes aspectos:\n",
        "\n",
        "1. Hacer pruebas con conjuntos de train, test y validation.\n",
        "\n",
        "2. Probar usar un **cuantil** en lugar del máximo de la variable de intentos fallidos, y de tiempo total.\n",
        "\n",
        "3. Probar combinar eso último con quitar o dejar los ejercicios con mucha deserción.\n",
        "\n",
        "4. Cambiar la variable categórica BINARIA “Hizo / No lo hizo”, por otra que sea **categórica ORDINAL** (no lo hizo, lo hizo mal, lo hizo regular, lo hizo bien). \n",
        "Notar: \n",
        "Como establecimos un supuesto respecto a los casos perdidos (asimilarlos al perfil desertor, o de “dificultad en la resolución”; y no a un perfil que se relaciona con saltear ejercicios porque son fáciles), podemos entonces ubicar los Nan en un  ordenamiento de status, por ejemplo: (Nans) = no lo hizo, (status Failed y Errored) = lo hizo mal, (status Passed with warning) = lo hizo regular , y (status Passed)  = lo hizo bien.\n"
      ],
      "id": "t5ptlFcEFFyo"
    },
    {
      "cell_type": "markdown",
      "metadata": {
        "id": "wTGLWtyhhqB5"
      },
      "source": [
        "# FIN"
      ],
      "id": "wTGLWtyhhqB5"
    }
  ]
}