CNN-Implementation.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "6eae8cb0",
   "metadata": {},
   "outputs": [],
   "source": [
    "# import necessary libraries\n",
    "import pickle\n",
    "import json\n",
    "import random\n",
    "import os.path\n",
    "\n",
    "from keras.models import load_model\n",
    "from keras.preprocessing.sequence import pad_sequences"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "id": "9561ae56",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[nltk_data] Downloading package stopwords to\n",
      "[nltk_data]     C:\\Users\\94773\\AppData\\Roaming\\nltk_data...\n",
      "[nltk_data]   Package stopwords is already up-to-date!\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "True"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "import nltk\n",
    "from nltk.corpus import stopwords\n",
    "from textblob import Word\n",
    "nltk.download('stopwords')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "id": "6bf5afde",
   "metadata": {},
   "outputs": [],
   "source": [
    "# tokenizer object loaded to map texts into sequences\n",
    "with open('tokenizer.pickle', 'rb') as handle:\n",
    "    tokenizer = pickle.load(handle)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "c41f7e56",
   "metadata": {},
   "outputs": [],
   "source": [
    "classes = [\"neutral\", \"happy\", \"sad\", \"love\", \"anger\"]\n",
    "\n",
    "# load trained model\n",
    "model = load_model(\"BESTCNNWeightsNLTP.h5\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "6afc5a31",
   "metadata": {},
   "outputs": [],
   "source": [
    "to_test = \"so happy\""
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "76d49b52",
   "metadata": {},
   "outputs": [],
   "source": [
    "stop = stopwords.words('english')\n",
    "\n",
    "to_test = to_test.replace('[^\\w\\s]',' ')\n",
    "\n",
    "to_test = ' '.join([word for word in to_test.split() if word not in stop])\n",
    "\n",
    "to_test = \" \".join([Word(word).lemmatize() for word in to_test.split()])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "e0b52284",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "happy\n"
     ]
    }
   ],
   "source": [
    "print(to_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "c44eb3d6",
   "metadata": {},
   "outputs": [],
   "source": [
    "to_test = [to_test]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "88f2a32d",
   "metadata": {},
   "outputs": [],
   "source": [
    "MAX_SEQUENCE_LENGTH = 30 # max length of text (words) including padding\n",
    "from keras.preprocessing.text import Tokenizer\n",
    "from keras_preprocessing.sequence import pad_sequences\n",
    "\n",
    "sequences_test = tokenizer.texts_to_sequences(to_test)\n",
    "data_int_t = pad_sequences(sequences_test, padding='pre', maxlen=(MAX_SEQUENCE_LENGTH-5))\n",
    "data_test = pad_sequences(data_int_t, padding='post', maxlen=(MAX_SEQUENCE_LENGTH))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "e54ba952",
   "metadata": {},
   "outputs": [],
   "source": [
    "y_prob = model.predict(data_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "id": "8ed98ce5",
   "metadata": {},
   "outputs": [],
   "source": [
    "pred_scores = y_prob[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "7b00705e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.34639525, 0.40660602, 0.0880124 , 0.15566762, 0.00331875],\n",
       "      dtype=float32)"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pred_scores"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "id": "519c7287",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.088012405"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pred_scores[2]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "id": "4439dff3",
   "metadata": {},
   "outputs": [],
   "source": [
    "for n, prediction in enumerate(y_prob):\n",
    "        pred_index = y_prob.argmax(axis=-1)[n]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "id": "7c61339d",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pred_index"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "70f175f3",
   "metadata": {},
   "outputs": [],
   "source": [
    "#Testing the model on 50, handpicked messages"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "id": "bb33dbae",
   "metadata": {},
   "outputs": [],
   "source": [
    "classes = [\"neutral\", \"happy\", \"sad\", \"love\", \"anger\"]\n",
    "\n",
    "# load trained model\n",
    "model = load_model(\"BESTCNNWeightsNLTP.h5\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "id": "e2954bae",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "\n",
    "df = pd.read_csv(\"Handpicked.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "id": "d0d5875a",
   "metadata": {},
   "outputs": [
    {
     "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>Text</th>\n",
       "      <th>Emotion</th>\n",
       "      <th>EmotionSVMLOGREG</th>\n",
       "      <th>EmotionsCNN</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>I don't feel like doing anything today, everyt...</td>\n",
       "      <td>sad</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>I just found out my grandmother passed away an...</td>\n",
       "      <td>sad</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>I feel so alone and isolated, like no one unde...</td>\n",
       "      <td>sad</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>I can't believe I lost my job, how am I going ...</td>\n",
       "      <td>sad</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>I miss my ex so much, it's hard to move on.</td>\n",
       "      <td>sad</td>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                                                Text Emotion  \\\n",
       "0  I don't feel like doing anything today, everyt...     sad   \n",
       "1  I just found out my grandmother passed away an...     sad   \n",
       "2  I feel so alone and isolated, like no one unde...     sad   \n",
       "3  I can't believe I lost my job, how am I going ...     sad   \n",
       "4        I miss my ex so much, it's hard to move on.     sad   \n",
       "\n",
       "   EmotionSVMLOGREG  EmotionsCNN  \n",
       "0                 4            2  \n",
       "1                 4            2  \n",
       "2                 4            2  \n",
       "3                 4            2  \n",
       "4                 4            2  "
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "id": "ab2c39e5",
   "metadata": {},
   "outputs": [],
   "source": [
    "actual_emotion = df['EmotionsCNN']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "id": "cc0d8394",
   "metadata": {},
   "outputs": [],
   "source": [
    "to_test = df['Text']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "id": "2ce942a3",
   "metadata": {},
   "outputs": [],
   "source": [
    "new_list = []\n",
    "for i in range(len(to_test)):\n",
    "    new_list.append(list(to_test[i]))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "id": "32639467",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[[\"I don't feel like doing anything today, everything seems pointless.\"],\n",
       " [\"I just found out my grandmother passed away and I can't stop crying.\"],\n",
       " ['I feel so alone and isolated, like no one understands me.'],\n",
       " [\"I can't believe I lost my job, how am I going to pay my bills?\"],\n",
       " [\"I miss my ex so much, it's hard to move on.\"],\n",
       " [\" I'm struggling with depression and it's hard to find motivation to get out of bed.\"],\n",
       " [\"I feel like I'm constantly disappointing everyone around me.\"],\n",
       " [\"I received some bad news from the doctor and I'm feeling really down.\"],\n",
       " [\"I'm so heartbroken, my pet just passed away.\"],\n",
       " [\"I feel like I'm stuck in a rut and I don't know how to break free.\"],\n",
       " [\"I can't believe they did that to me, I'm so furious!\"],\n",
       " [\"I'm sick and tired of dealing with this nonsense, I'm about to snap.\"],\n",
       " [\"Why does everything have to be so difficult? I'm getting really angry.\"],\n",
       " [\"I feel so disrespected and undervalued, it's making me angry.\"],\n",
       " [\"I can't stand how unfair this situation is, it's making me really angry.\"],\n",
       " [\"I'm so angry at myself for letting things get this far.\"],\n",
       " [\"I'm about to lose my temper if this keeps happening.\"],\n",
       " [\"I can't believe they had the audacity to say that to me, I'm so angry.\"],\n",
       " [\"I'm so frustrated with this situation, it's making me see red.\"],\n",
       " [\"I'm not going to stand for this anymore, I'm feeling really angry and fed up.\"],\n",
       " [\"I'm looking forward to trying out that new restaurant in town.\"],\n",
       " ['I have a meeting at 2 pm tomorrow, so I need to finish my work before then.'],\n",
       " ['The weather today is sunny with a high of 75 degrees.'],\n",
       " [\"I'm currently reading a book about World War II.\"],\n",
       " ['I need to go grocery shopping and do laundry this weekend.'],\n",
       " [\"I'm thinking about going on a vacation next month, but I haven't decided where yet.\"],\n",
       " [\"I'm enjoying this cup of coffee while I catch up on some work.\"],\n",
       " [\"My favorite color is blue, what's yours?\"],\n",
       " [\"I'm learning how to play the guitar in my free time.\"],\n",
       " [\"I'm planning to go for a run after work to clear my mind.\"],\n",
       " ['I love spending time with my mother'],\n",
       " ['You make my day brighter and I love it so much '],\n",
       " ['I think I met my soulmate'],\n",
       " ['I am head over heels for my current partner'],\n",
       " ['I love my family so much'],\n",
       " ['OMG I LOVE STEPH CURRY'],\n",
       " [\"There's nothing I wont do for the love of my life\"],\n",
       " ['I love how we can just be ourselves around each other, you complete me.'],\n",
       " ['I appreciate everything you do for me, I love you more every day.'],\n",
       " [\"I can't imagine my life without you, you mean everything to me.\"],\n",
       " [\"I'm so happy that I got the job I wanted!\"],\n",
       " ['Today is a beautiful day, I feel happy just being alive.'],\n",
       " [\"I'm thrilled to be going on vacation next week!\"],\n",
       " [\"I just found out that I'm going to be a parent, and I'm overjoyed!\"],\n",
       " [\"I'm so grateful for my family and friends, they make me incredibly happy.\"],\n",
       " [\"I'm feeling so happy and content right now, life is good.\"],\n",
       " ['I love spending time doing things that make me happy, like reading or going for walks.'],\n",
       " [\"I'm really proud of myself for accomplishing my goals, it makes me happy to see my hard work pay off.\"],\n",
       " ['I love waking up early on weekends and enjoying a quiet morning with a cup of coffee, it makes me so happy.'],\n",
       " [\"I'm so happy to be surrounded by positivity and good vibes, it makes life so much better.\"]]"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "new_list"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "id": "1af5f7ba",
   "metadata": {},
   "outputs": [],
   "source": [
    "MAX_SEQUENCE_LENGTH = 30 # max length of text (words) including padding\n",
    "from keras.preprocessing.text import Tokenizer\n",
    "from keras_preprocessing.sequence import pad_sequences"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6cebe969",
   "metadata": {},
   "outputs": [],
   "source": [
    "sequences_test = tokenizer.texts_to_sequences(to_test)\n",
    "data_int_t = pad_sequences(sequences_test, padding='pre', maxlen=(MAX_SEQUENCE_LENGTH-5))\n",
    "data_test = pad_sequences(data_int_t, padding='post', maxlen=(MAX_SEQUENCE_LENGTH))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "id": "87f0840f",
   "metadata": {},
   "outputs": [],
   "source": [
    "final_list = []\n",
    "\n",
    "for i in new_list:\n",
    "    sequences_test = tokenizer.texts_to_sequences(i)\n",
    "    data_int_t = pad_sequences(sequences_test, padding='pre', maxlen=(MAX_SEQUENCE_LENGTH-5))\n",
    "    data_test = pad_sequences(data_int_t, padding='post', maxlen=(MAX_SEQUENCE_LENGTH))\n",
    "    final_list.append(data_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 31,
   "id": "17847601",
   "metadata": {},
   "outputs": [],
   "source": [
    "y_prob = model.predict(final_list[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 33,
   "id": "d228a1ed",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "50"
      ]
     },
     "execution_count": 33,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(final_list)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 32,
   "id": "c7527ffa",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "2"
      ]
     },
     "execution_count": 32,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y_prob.argmax()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 37,
   "id": "791940e8",
   "metadata": {},
   "outputs": [],
   "source": [
    "pred = []"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 38,
   "id": "8d13e4f5",
   "metadata": {},
   "outputs": [],
   "source": [
    "for i in range(len(final_list)):\n",
    "    pred.append(model.predict(final_list[i]).argmax())"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 39,
   "id": "c9a0faf9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[2,\n",
       " 2,\n",
       " 2,\n",
       " 2,\n",
       " 2,\n",
       " 2,\n",
       " 2,\n",
       " 2,\n",
       " 2,\n",
       " 2,\n",
       " 2,\n",
       " 2,\n",
       " 4,\n",
       " 4,\n",
       " 4,\n",
       " 4,\n",
       " 2,\n",
       " 4,\n",
       " 2,\n",
       " 4,\n",
       " 0,\n",
       " 2,\n",
       " 1,\n",
       " 2,\n",
       " 2,\n",
       " 2,\n",
       " 1,\n",
       " 1,\n",
       " 1,\n",
       " 2,\n",
       " 3,\n",
       " 3,\n",
       " 1,\n",
       " 2,\n",
       " 3,\n",
       " 3,\n",
       " 3,\n",
       " 3,\n",
       " 3,\n",
       " 3,\n",
       " 2,\n",
       " 3,\n",
       " 1,\n",
       " 1,\n",
       " 3,\n",
       " 3,\n",
       " 3,\n",
       " 1,\n",
       " 3,\n",
       " 1]"
      ]
     },
     "execution_count": 39,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pred"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 40,
   "id": "bcf6a862",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import classification_report\n",
    "import seaborn as sn\n",
    "from sklearn.metrics import confusion_matrix\n",
    "import matplotlib.pyplot as plt\n",
    "import numpy\n",
    "from sklearn import metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 41,
   "id": "f8a7b68e",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "              precision    recall  f1-score   support\n",
      "\n",
      "           0       1.00      0.10      0.18        10\n",
      "           1       0.44      0.40      0.42        10\n",
      "           2       0.48      1.00      0.65        10\n",
      "           3       0.62      0.80      0.70        10\n",
      "           4       1.00      0.60      0.75        10\n",
      "\n",
      "    accuracy                           0.58        50\n",
      "   macro avg       0.71      0.58      0.54        50\n",
      "weighted avg       0.71      0.58      0.54        50\n",
      "\n"
     ]
    }
   ],
   "source": [
    "print(classification_report(actual_emotion, pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 42,
   "id": "099084e4",
   "metadata": {},
   "outputs": [],
   "source": [
    "x = confusion_matrix(actual_emotion, pred)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 43,
   "id": "00859849",
   "metadata": {},
   "outputs": [],
   "source": [
    "classes = [\"Neutral\", \"Happy\", \"Sad\", \"Love\", \"Anger\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "27582221",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<AxesSubplot:>"
      ]
     },
     "execution_count": 44,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 1080x720 with 2 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "df_cm = pd.DataFrame(x, index = [i for i in classes],\n",
    "                  columns = [i for i in classes])\n",
    "plt.figure(figsize = (15,10))\n",
    "\n",
    "sn.heatmap(df_cm, annot=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "ab2e62dc",
   "metadata": {},
   "outputs": [],
   "source": [
    "#Testing the model on 8 different messages"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "id": "e1b5cea8",
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "[nltk_data] Downloading package stopwords to\n",
      "[nltk_data]     C:\\Users\\94773\\AppData\\Roaming\\nltk_data...\n",
      "[nltk_data]   Package stopwords is already up-to-date!\n",
      "[nltk_data] Downloading package wordnet to\n",
      "[nltk_data]     C:\\Users\\94773\\AppData\\Roaming\\nltk_data...\n",
      "[nltk_data]   Package wordnet is already up-to-date!\n"
     ]
    }
   ],
   "source": [
    "import pickle\n",
    "import nltk\n",
    "nltk.download('stopwords')\n",
    "nltk.download('wordnet')\n",
    "\n",
    "from nltk.corpus import stopwords\n",
    "from textblob import Word\n",
    "\n",
    "stop = stopwords.words('english')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "id": "fb7bdd8a",
   "metadata": {},
   "outputs": [],
   "source": [
    "# tokenizer object loaded to map texts into sequences\n",
    "with open('tokenizer.pickle', 'rb') as handle:\n",
    "    tokenizer = pickle.load(handle)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "id": "3b5cab3d",
   "metadata": {},
   "outputs": [],
   "source": [
    "from keras.models import load_model\n",
    "from keras.preprocessing.sequence import pad_sequences"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 30,
   "id": "706e3738",
   "metadata": {},
   "outputs": [],
   "source": [
    "model = load_model(\"CNNWeights.h5\")\n",
    "stop = stopwords.words('english')\n",
    "MAX_SEQUENCE_LENGTH = 30"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 44,
   "id": "42b6ce18",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0.18423396 0.36593562 0.1986597  0.24593389 0.00523675]\n"
     ]
    }
   ],
   "source": [
    "message = \"im happy to meet my parents soon!\"\n",
    "message = ' '.join([word for word in message.split() if word not in stop])\n",
    "message = \" \".join([Word(word).lemmatize() for word in message.split()])\n",
    "message = [message]\n",
    "\n",
    "sequences = tokenizer.texts_to_sequences(message)\n",
    "pre_padding = pad_sequences(sequences, padding='pre', maxlen=(MAX_SEQUENCE_LENGTH - 5))\n",
    "post_padding = pad_sequences(pre_padding, padding='post', maxlen=MAX_SEQUENCE_LENGTH)\n",
    "\n",
    "y_prob = model.predict(post_padding)\n",
    "print(y_prob[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "8e08e324",
   "metadata": {},
   "outputs": [],
   "source": [
    "#[\"neutral\", \"happy\", \"sad\", \"love\", \"anger\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "d8fd2a15",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3 (ipykernel)",
   "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": "3.9.12"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}