3-In-Memory-Tables.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Querying tables"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "> Objectives:\n",
    "> * Compare queries of tabular data for **in-memory** containers\n",
    "> * Compare sizes and times for those"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [1] used 0.023 MiB RAM in 0.000s, peaked 0.000 MiB above current, total RAM usage 30.141 MiB\n"
     ]
    }
   ],
   "source": [
    "from ipython_memwatcher import MemWatcher\n",
    "mw = MemWatcher()\n",
    "mw.start_watching_memory()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [2] used 0.027 MiB RAM in 0.001s, peaked 0.000 MiB above current, total RAM usage 30.168 MiB\n"
     ]
    }
   ],
   "source": [
    "import os\n",
    "dset = 'movielens-1m'\n",
    "fdata = os.path.join(dset, 'ratings.dat.gz')\n",
    "fitem = os.path.join(dset, 'movies.dat')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [3] used 71.434 MiB RAM in 0.472s, peaked 24.598 MiB above current, total RAM usage 101.602 MiB\n"
     ]
    }
   ],
   "source": [
    "import pandas as pd\n",
    "# pass in column names for each CSV\n",
    "r_cols = ['user_id', 'movie_id', 'rating', 'unix_timestamp']\n",
    "ratings = pd.read_csv(fdata, sep=';', names=r_cols, compression='gzip')\n",
    "\n",
    "m_cols = ['movie_id', 'title', 'genres']\n",
    "movies = pd.read_csv(fitem, sep=';', names=m_cols,\n",
    "                     dtype={'title': object, 'genres': object})"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [4] used 62.375 MiB RAM in 0.132s, peaked 30.871 MiB above current, total RAM usage 163.977 MiB\n"
     ]
    }
   ],
   "source": [
    "lens = pd.merge(movies, ratings)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "62.375"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [5] used 0.043 MiB RAM in 0.005s, peaked 0.000 MiB above current, total RAM usage 164.020 MiB\n"
     ]
    }
   ],
   "source": [
    "size_pandas = mw.measurements.memory_delta\n",
    "size_pandas"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "   movie_id             title                       genres  user_id  rating  \\\n",
      "0         1  Toy Story (1995)  Animation|Children's|Comedy        1       5   \n",
      "1         1  Toy Story (1995)  Animation|Children's|Comedy        6       4   \n",
      "2         1  Toy Story (1995)  Animation|Children's|Comedy        8       4   \n",
      "3         1  Toy Story (1995)  Animation|Children's|Comedy        9       5   \n",
      "4         1  Toy Story (1995)  Animation|Children's|Comedy       10       5   \n",
      "5         1  Toy Story (1995)  Animation|Children's|Comedy       18       4   \n",
      "6         1  Toy Story (1995)  Animation|Children's|Comedy       19       5   \n",
      "7         1  Toy Story (1995)  Animation|Children's|Comedy       21       3   \n",
      "8         1  Toy Story (1995)  Animation|Children's|Comedy       23       4   \n",
      "9         1  Toy Story (1995)  Animation|Children's|Comedy       26       3   \n",
      "\n",
      "   unix_timestamp  \n",
      "0       978824268  \n",
      "1       978237008  \n",
      "2       978233496  \n",
      "3       978225952  \n",
      "4       978226474  \n",
      "5       978154768  \n",
      "6       978555994  \n",
      "7       978139347  \n",
      "8       978463614  \n",
      "9       978130703  \n",
      "<class 'pandas.core.frame.DataFrame'>\n",
      "Int64Index: 1000209 entries, 0 to 1000208\n",
      "Data columns (total 6 columns):\n",
      "movie_id          1000209 non-null int64\n",
      "title             1000209 non-null object\n",
      "genres            1000209 non-null object\n",
      "user_id           1000209 non-null int64\n",
      "rating            1000209 non-null int64\n",
      "unix_timestamp    1000209 non-null int64\n",
      "dtypes: int64(4), object(2)\n",
      "memory usage: 53.4+ MB\n",
      "In [6] used 14.320 MiB RAM in 0.150s, peaked 0.000 MiB above current, total RAM usage 178.340 MiB\n"
     ]
    }
   ],
   "source": [
    "print(lens[:10])\n",
    "lens.info()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "45.785934448242188"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [7] used 0.031 MiB RAM in 0.002s, peaked 0.000 MiB above current, total RAM usage 178.371 MiB\n"
     ]
    }
   ],
   "source": [
    "size_pandas2 = lens.memory_usage().sum(index=True) / 2**20.\n",
    "size_pandas2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "10 loops, best of 3: 31.3 ms per loop\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "5121      75\n",
       "5164    3842\n",
       "5187    6031\n",
       "Name: user_id, dtype: int64"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [8] used 0.215 MiB RAM in 1.337s, peaked 0.000 MiB above current, total RAM usage 178.586 MiB\n"
     ]
    }
   ],
   "source": [
    "result = lens.query(\"(title == 'Tom and Huck (1995)') & (rating == 5)\")['user_id']\n",
    "t = %timeit -o lens.query(\"(title == 'Tom and Huck (1995)') & (rating == 5)\")['user_id']\n",
    "result"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [9] used 0.004 MiB RAM in 0.001s, peaked 0.000 MiB above current, total RAM usage 178.590 MiB\n"
     ]
    }
   ],
   "source": [
    "qtime_pandas = t.best"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Use a compressed in-memory container via `bcolz`"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=\n",
      "bcolz version:     0.12.1.dev2\n",
      "NumPy version:     1.10.1\n",
      "Blosc version:     1.4.1 ($Date:: 2014-07-08 #$)\n",
      "Blosc compressors: ['blosclz', 'lz4', 'lz4hc', 'snappy', 'zlib']\n",
      "Numexpr version:   2.4.4\n",
      "Python version:    2.7.10 |Continuum Analytics, Inc.| (default, Oct 19 2015, 18:04:42) \n",
      "[GCC 4.4.7 20120313 (Red Hat 4.4.7-1)]\n",
      "Platform:          linux2-x86_64\n",
      "Byte-ordering:     little\n",
      "Detected cores:    8\n",
      "-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "8"
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [10] used 0.688 MiB RAM in 0.013s, peaked 0.000 MiB above current, total RAM usage 179.277 MiB\n"
     ]
    }
   ],
   "source": [
    "import bcolz\n",
    "bcolz.print_versions()\n",
    "bcolz.defaults.cparams['cname'] = 'lz4'\n",
    "bcolz.defaults.cparams['clevel'] = 5\n",
    "bcolz.set_nthreads(4)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [11] used 0.480 MiB RAM in 0.174s, peaked 0.652 MiB above current, total RAM usage 179.758 MiB\n"
     ]
    }
   ],
   "source": [
    "zlens = bcolz.ctable.fromdataframe(lens)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.48046875"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [12] used 0.020 MiB RAM in 0.002s, peaked 0.000 MiB above current, total RAM usage 179.777 MiB\n"
     ]
    }
   ],
   "source": [
    "size_bcolz = mw.measurements.memory_delta\n",
    "size_bcolz"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Sometimes the memory reported as used is too much biased (don't know why), so let's repeat the operation, but using a different container:"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [13] used 0.059 MiB RAM in 0.173s, peaked 0.945 MiB above current, total RAM usage 179.836 MiB\n"
     ]
    }
   ],
   "source": [
    "zlens2 = bcolz.ctable.fromdataframe(lens)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "7.894172668457031"
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [14] used 0.020 MiB RAM in 0.002s, peaked 0.000 MiB above current, total RAM usage 179.855 MiB\n"
     ]
    }
   ],
   "source": [
    "size_bcolz2 = zlens2.cbytes / 2**20.\n",
    "size_bcolz2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "ctable((1000209,), [('movie_id', '<i8'), ('title', 'S82'), ('genres', 'S47'), ('user_id', '<i8'), ('rating', '<i8'), ('unix_timestamp', '<i8')])\n",
       "  nbytes: 153.57 MB; cbytes: 7.89 MB; ratio: 19.45\n",
       "  cparams := cparams(clevel=5, shuffle=True, cname='lz4')\n",
       "[(1, 'Toy Story (1995)', \"Animation|Children's|Comedy\", 1, 5, 978824268)\n",
       " (1, 'Toy Story (1995)', \"Animation|Children's|Comedy\", 6, 4, 978237008)\n",
       " (1, 'Toy Story (1995)', \"Animation|Children's|Comedy\", 8, 4, 978233496)\n",
       " ...,\n",
       " (3952, 'Contender, The (2000)', 'Drama|Thriller', 5837, 4, 1011902656)\n",
       " (3952, 'Contender, The (2000)', 'Drama|Thriller', 5927, 1, 979852537)\n",
       " (3952, 'Contender, The (2000)', 'Drama|Thriller', 5998, 4, 1001781044)]"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [15] used 0.012 MiB RAM in 0.004s, peaked 0.000 MiB above current, total RAM usage 179.867 MiB\n"
     ]
    }
   ],
   "source": [
    "zlens"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "129.8211382113821"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [16] used 0.000 MiB RAM in 0.001s, peaked 0.000 MiB above current, total RAM usage 179.867 MiB\n"
     ]
    }
   ],
   "source": [
    "size_pandas / size_bcolz"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We can see that the space taken by a bcolz container is more than 100x smaller (!) than a pandas one."
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Excercise\n",
    "\n",
    "Why do you think that the number of uncompressed bytes (nbytes) that the ctable reports is 3x more than pandas (153 MB vs 54 MB)?\n",
    "\n",
    "*Hint:* Pandas stores the string columns in NumPy containers with 'object' dtype whereas bcolz uses the equivalent to NumPy's 'string' objects."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "100 loops, best of 3: 18.8 ms per loop\n",
      "In [17] used 3.152 MiB RAM in 7.869s, peaked 0.000 MiB above current, total RAM usage 183.020 MiB\n"
     ]
    }
   ],
   "source": [
    "# Querying a bcolz dataset\n",
    "resultz = [(r.nrow__, r.user_id) for r in zlens.where(\"(title == b'Tom and Huck (1995)') & (rating == 5)\", outcols=['nrow__', 'user_id'])]\n",
    "t = %timeit -o [(r.nrow__, r.user_id) for r in zlens.where(\"(title == b'Tom and Huck (1995)') & (rating == 5)\", outcols=['nrow__', 'user_id'])]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [18] used 0.020 MiB RAM in 0.001s, peaked 0.000 MiB above current, total RAM usage 183.039 MiB\n"
     ]
    }
   ],
   "source": [
    "qtime_bcolz = t.best"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.66517625278669"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [19] used 0.000 MiB RAM in 0.001s, peaked 0.000 MiB above current, total RAM usage 183.039 MiB\n"
     ]
    }
   ],
   "source": [
    "qtime_pandas / qtime_bcolz"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "We see that bcolz containers, besides being in compressed state, provide ~1.8x faster query times than pandas"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "('results with pandas Dataframe:', 5121      75\n",
      "5164    3842\n",
      "5187    6031\n",
      "Name: user_id, dtype: int64)\n",
      "('results with bcolz ctable:', [(5121, 75), (5164, 3842), (5187, 6031)])\n",
      "In [20] used 0.004 MiB RAM in 0.002s, peaked 0.000 MiB above current, total RAM usage 183.043 MiB\n"
     ]
    }
   ],
   "source": [
    "# Sanity check (always check for your results!)\n",
    "print(\"results with pandas Dataframe:\", result)\n",
    "print(\"results with bcolz ctable:\", resultz)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {
    "collapsed": true
   },
   "source": [
    "## Using structured NumPy arrays"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [21] used 53.418 MiB RAM in 0.436s, peaked 0.000 MiB above current, total RAM usage 236.461 MiB\n"
     ]
    }
   ],
   "source": [
    "nalens = lens.to_records()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "53.41796875"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [22] used 0.023 MiB RAM in 0.004s, peaked 0.000 MiB above current, total RAM usage 236.484 MiB\n"
     ]
    }
   ],
   "source": [
    "size_numpy = mw.measurements.memory_delta\n",
    "size_numpy"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "rec.array([ (0, 1, 'Toy Story (1995)', \"Animation|Children's|Comedy\", 1, 5, 978824268),\n",
       " (1, 1, 'Toy Story (1995)', \"Animation|Children's|Comedy\", 6, 4, 978237008),\n",
       " (2, 1, 'Toy Story (1995)', \"Animation|Children's|Comedy\", 8, 4, 978233496),\n",
       " ...,\n",
       " (1000206, 3952, 'Contender, The (2000)', 'Drama|Thriller', 5837, 4, 1011902656),\n",
       " (1000207, 3952, 'Contender, The (2000)', 'Drama|Thriller', 5927, 1, 979852537),\n",
       " (1000208, 3952, 'Contender, The (2000)', 'Drama|Thriller', 5998, 4, 1001781044)], \n",
       "          dtype=[('index', '<i8'), ('movie_id', '<i8'), ('title', '|O'), ('genres', '|O'), ('user_id', '<i8'), ('rating', '<i8'), ('unix_timestamp', '<i8')])"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [23] used 0.004 MiB RAM in 0.005s, peaked 0.000 MiB above current, total RAM usage 236.488 MiB\n"
     ]
    }
   ],
   "source": [
    "nalens"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "53.417"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [24] used 0.000 MiB RAM in 0.004s, peaked 0.000 MiB above current, total RAM usage 236.488 MiB\n"
     ]
    }
   ],
   "source": [
    "size_numpy2 = round((nalens.size * nalens.dtype.itemsize) / 2**20., 3)\n",
    "size_numpy2"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "100 loops, best of 3: 13.6 ms per loop\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "rec.array([ (5121, 8, 'Tom and Huck (1995)', \"Adventure|Children's\", 75, 5, 977851520),\n",
       " (5164, 8, 'Tom and Huck (1995)', \"Adventure|Children's\", 3842, 5, 967986151),\n",
       " (5187, 8, 'Tom and Huck (1995)', \"Adventure|Children's\", 6031, 5, 956718223)], \n",
       "          dtype=[('index', '<i8'), ('movie_id', '<i8'), ('title', '|O'), ('genres', '|O'), ('user_id', '<i8'), ('rating', '<i8'), ('unix_timestamp', '<i8')])"
      ]
     },
     "execution_count": 25,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [25] used 0.023 MiB RAM in 5.707s, peaked 0.000 MiB above current, total RAM usage 236.512 MiB\n"
     ]
    }
   ],
   "source": [
    "resultna = nalens[(nalens['title'] == 'Tom and Huck (1995)') & (nalens['rating'] == 5)]\n",
    "t = %timeit -o nalens[(nalens['title'] == 'Tom and Huck (1995)') & (nalens['rating'] == 5)]\n",
    "resultna"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "Again, NumPy works the fastest for in-memory data containers, while memory consumption is close to pandas."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [26] used 0.023 MiB RAM in 0.001s, peaked 0.000 MiB above current, total RAM usage 236.535 MiB\n"
     ]
    }
   ],
   "source": [
    "qtime_numpy = t.best"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Performance comparsion"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {
    "collapsed": false
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [27] used 5.574 MiB RAM in 0.069s, peaked 0.004 MiB above current, total RAM usage 242.109 MiB\n"
     ]
    }
   ],
   "source": [
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f6508ebcd10>"
      ]
     },
     "execution_count": 28,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/francesc/miniconda/lib/python2.7/site-packages/matplotlib/font_manager.py:1287: UserWarning: findfont: Font family [u'monospace'] not found. Falling back to Bitstream Vera Sans\n",
      "  (prop.get_family(), self.defaultFamily[fontext]))\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA1YAAAFKCAYAAAAe186iAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\nhxisAAAAAMChP+TwRudsKplIAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f6508ebc350>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [28] used 4.715 MiB RAM in 0.225s, peaked 0.000 MiB above current, total RAM usage 246.824 MiB\n"
     ]
    }
   ],
   "source": [
    "# Query times\n",
    "labels = [\"pandas\", \"bcolz\", \"numpy (recarray)\"]\n",
    "df = pd.DataFrame({'time (sec)': [qtime_pandas, qtime_bcolz, qtime_numpy]}, index=labels)\n",
    "pd.options.display.mpl_style = 'default'\n",
    "df.plot(kind='barh', figsize=(12,5), fontsize=16, title=\"Query times for MovieLens 1m (in-memory)\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Size comparison"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {
    "collapsed": false,
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x7f6508e8f510>"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAA0kAAAFKCAYAAADIchTHAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n2YjKAAAAAElFTkSuQmCC\n",
      "text/plain": [
       "<matplotlib.figure.Figure at 0x7f65043e6dd0>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "In [29] used 3.609 MiB RAM in 0.180s, peaked 0.000 MiB above current, total RAM usage 250.434 MiB\n"
     ]
    }
   ],
   "source": [
    "# Container sizes\n",
    "df = pd.DataFrame({'size (MB)': [size_pandas, size_bcolz, size_numpy]}, index=labels)\n",
    "pd.options.display.mpl_style = 'default'\n",
    "df.plot(kind='barh', figsize=(12,5), fontsize=16, title=\"Container sizes for MovieLens 1m\")"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Rules of thumb for querying in-memory tabular datasets\n",
    "\n",
    "* Choose pure NumPy recarrays if you need the fastest speed\n",
    "* Choose bcolz ctables if you need to store lots of data in limited memory and not want to loose too much speed\n",
    "* Choose pandas if what you need is rich functionality on top of your containers (at the penalty of some speed)"
   ]
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 2",
   "language": "python",
   "name": "python2"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 2
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython2",
   "version": "2.7.10"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 0
}