05.05_Feature_Hashing.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import json"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(10000, 8)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# Load the first 10000 reviews\n",
    "f = open('data/yelp/v6/yelp_dataset_challenge_academic_dataset/yelp_academic_dataset_review.json')\n",
    "js = []\n",
    "for i in range(10000):\n",
    "    js.append(json.loads(f.readline()))\n",
    "f.close()\n",
    "review_df = pd.DataFrame(js)\n",
    "review_df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "528\n"
     ]
    }
   ],
   "source": [
    "# we will define m as equal to the unique number of business_id\n",
    "m = len(review_df.business_id.unique())\n",
    "print(m)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.feature_extraction import FeatureHasher"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "h = FeatureHasher(n_features=m, input_type='string')\n",
    "f = h.transform(review_df['business_id'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 28,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Our pandas Series, in bytes:  790104\n",
      "Our hashed numpy array, in bytes:  56\n"
     ]
    }
   ],
   "source": [
    "# We can see how this will make a difference in the future by looking at the size of each\n",
    "from sys import getsizeof\n",
    "\n",
    "print('Our pandas Series, in bytes: ', getsizeof(review_df['business_id']))\n",
    "print('Our hashed numpy array, in bytes: ', getsizeof(f))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['vcNAWiLM4dR7D2nwwJ7nCA',\n",
       " 'UsFtqoBl7naz8AVUBZMjQQ',\n",
       " 'cE27W9VPgO88Qxe4ol6y_g',\n",
       " 'HZdLhv6COCleJMo7nPl-RA',\n",
       " 'mVHrayjG3uZ_RLHkLj-AMg']"
      ]
     },
     "execution_count": 19,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "review_df['business_id'].unique().tolist()[0:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 0.,  0.,  0., ...,  0.,  0.,  0.],\n",
       "       [ 0.,  0.,  0., ...,  0.,  0.,  0.],\n",
       "       [ 0.,  0.,  0., ...,  0.,  0.,  0.],\n",
       "       ..., \n",
       "       [ 0.,  0.,  0., ...,  0.,  0.,  0.],\n",
       "       [ 0.,  0.,  0., ...,  0.,  0.,  0.],\n",
       "       [ 0.,  0.,  0., ...,  0.,  0.,  0.]])"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "f.toarray()"
   ]
  }
 ],
 "metadata": {
  "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": "3.6.3"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}