recommendation_task.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 前處理"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import scipy.sparse as sparse\n",
    "from scipy.sparse.linalg import spsolve\n",
    "import pandas as pd\n",
    "import numpy as np"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "retail_data = pd.read_csv(\"rs.csv\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>Unnamed: 0</th>\n",
       "      <th>user</th>\n",
       "      <th>item</th>\n",
       "      <th>qty</th>\n",
       "      <th>datetime</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>32011003635952</td>\n",
       "      <td>16198</td>\n",
       "      <td>1</td>\n",
       "      <td>2014-01-01 00:39:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>32009100468450</td>\n",
       "      <td>18107</td>\n",
       "      <td>4</td>\n",
       "      <td>2014-01-01 01:07:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>32013007873699</td>\n",
       "      <td>13612</td>\n",
       "      <td>2</td>\n",
       "      <td>2014-01-01 07:00:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>32011004803503</td>\n",
       "      <td>42496</td>\n",
       "      <td>1</td>\n",
       "      <td>2014-01-01 07:06:00</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>32011004221857</td>\n",
       "      <td>8875</td>\n",
       "      <td>4</td>\n",
       "      <td>2014-01-01 07:09:00</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   Unnamed: 0            user   item  qty             datetime\n",
       "0           0  32011003635952  16198    1  2014-01-01 00:39:00\n",
       "1           1  32009100468450  18107    4  2014-01-01 01:07:00\n",
       "2           2  32013007873699  13612    2  2014-01-01 07:00:00\n",
       "3           3  32011004803503  42496    1  2014-01-01 07:06:00\n",
       "4           4  32011004221857   8875    4  2014-01-01 07:09:00"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "retail_data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "item_lookup = retail_data[['item']].drop_duplicates()# 物品清單(無重複)\n",
    "item_lookup['item'] = item_lookup.item"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>item</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>16198</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>18107</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>13612</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>42496</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>8875</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    item\n",
       "0  16198\n",
       "1  18107\n",
       "2  13612\n",
       "3  42496\n",
       "4   8875"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "item_lookup.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/opt/conda/lib/python3.6/site-packages/pandas/core/indexing.py:141: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  self._setitem_with_indexer(indexer, value)\n"
     ]
    }
   ],
   "source": [
    "retail_data['user'] = retail_data.user.astype(\"str\")\n",
    "retail_data = retail_data[[ 'user','item', 'qty',]] # 只保留我們想要的值\n",
    "grouped_cleaned = retail_data.groupby(['user', 'item']).sum().reset_index() # 將所有購物紀錄結合起來\n",
    "grouped_cleaned.qty.loc[grouped_cleaned.qty == 0] = 1 # 將 0值轉成 1\n",
    "grouped_purchased = grouped_cleaned.query('qty > 0') # 只保留正值"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>user</th>\n",
       "      <th>item</th>\n",
       "      <th>qty</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>32009000000620</td>\n",
       "      <td>6095</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>32009000000620</td>\n",
       "      <td>6937</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>32009000000620</td>\n",
       "      <td>7650</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>32009000000620</td>\n",
       "      <td>8838</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>32009000000620</td>\n",
       "      <td>8841</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "             user  item  qty\n",
       "0  32009000000620  6095    1\n",
       "1  32009000000620  6937    1\n",
       "2  32009000000620  7650    2\n",
       "3  32009000000620  8838    4\n",
       "4  32009000000620  8841    5"
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "grouped_purchased.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "customers = list(np.sort(grouped_purchased.user.unique())) #顧客清單\n",
    "products = list(grouped_purchased.item.unique()) # 產品清單\n",
    "quantity = list(grouped_purchased.qty) # 溝買紀錄\n",
    "#\n",
    "rows = grouped_purchased.user.astype('category', categories = customers).cat.codes \n",
    "cols = grouped_purchased.item.astype('category', categories = products).cat.codes \n",
    "purchases_sparse = sparse.csr_matrix((quantity, (rows, cols)), shape=(len(customers), len(products)))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "95.12475"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "matrix_size = purchases_sparse.shape[0]*purchases_sparse.shape[1] # Number of possible interactions in the matrix\n",
    "num_purchases = len(purchases_sparse.nonzero()[0]) # Number of items interacted with\n",
    "sparsity = 100*(1 - (num_purchases/matrix_size))\n",
    "sparsity\n",
    "#95.12% 關聯矩陣為稀疏矩陣"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 分成訓練集和測試集"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "import random"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def make_train(ratings, pct_test = 0.2):\n",
    "    test_set = ratings.copy() # Make a copy of the original set to be the test set. \n",
    "    test_set[test_set != 0] = 1 # Store the test set as a binary preference matrix\n",
    "    training_set = ratings.copy() # Make a copy of the original data we can alter as our training set. \n",
    "    nonzero_inds = training_set.nonzero() \n",
    "    nonzero_pairs = list(zip(nonzero_inds[0], nonzero_inds[1])) # 利用zip結合user,item\n",
    "    random.seed(0) # 設定seed\n",
    "    num_samples = int(np.ceil(pct_test*len(nonzero_pairs))) # 四捨五入\n",
    "    samples = random.sample(nonzero_pairs, num_samples) \n",
    "    user_inds = [index[0] for index in samples] # Get the user row indices\n",
    "    item_inds = [index[1] for index in samples] # Get the item column indices\n",
    "    training_set[user_inds, item_inds] = 0 \n",
    "    training_set.eliminate_zeros() \n",
    "    return training_set, test_set, list(set(user_inds)) "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "product_train, product_test, product_users_altered = make_train(purchases_sparse, pct_test = 0.2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "This method is deprecated. Please use the AlternatingLeastSquares class instead\n",
      "WARNING:root:OpenBLAS detected. Its highly recommend to set the environment variable 'export OPENBLAS_NUM_THREADS=1' to disable its internal multithreading\n"
     ]
    }
   ],
   "source": [
    "## ALS 找出 user及 item 的 向量\n",
    "import implicit\n",
    "alpha = 15\n",
    "user_vecs, item_vecs = implicit.alternating_least_squares((product_train*alpha).astype('double'), \n",
    "                                                          factors=20, \n",
    "                                                          regularization = 0.1, \n",
    "                                                         iterations = 50)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 測AUC"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn import metrics"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def auc_score(predictions, test):\n",
    "    fpr, tpr, thresholds = metrics.roc_curve(test, predictions)\n",
    "    return metrics.auc(fpr, tpr)   "
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "def calc_mean_auc(training_set, altered_users, predictions, test_set):\n",
    "    store_auc = [] # An empty list to store the AUC for each user that had an item removed from the training set\n",
    "    popularity_auc = [] # To store popular AUC scores\n",
    "    pop_items = np.array(test_set.sum(axis = 0)).reshape(-1) # Get sum of item iteractions to find most popular\n",
    "    item_vecs = predictions[1]\n",
    "    for user in altered_users: # Iterate through each user that had an item altered\n",
    "        training_row = training_set[user,:].toarray().reshape(-1) # Get the training set row\n",
    "        zero_inds = np.where(training_row == 0) # Find where the interaction had not yet occurred\n",
    "        # Get the predicted values based on our user/item vectors\n",
    "        user_vec = predictions[0][user,:]\n",
    "        pred = user_vec.dot(item_vecs.T).toarray()[0,zero_inds].reshape(-1)\n",
    "        # Get only the items that were originally zero\n",
    "        # Select all ratings from the MF prediction for this user that originally had no iteraction\n",
    "        actual = test_set[user,:].toarray()[0,zero_inds].reshape(-1) \n",
    "        # Select the binarized yes/no interaction pairs from the original full data\n",
    "        # that align with the same pairs in training \n",
    "        pop = pop_items[zero_inds] # Get the item popularity for our chosen items\n",
    "        store_auc.append(auc_score(pred, actual)) # Calculate AUC for the given user and store\n",
    "        popularity_auc.append(auc_score(pop, actual)) # Calculate AUC using most popular and score\n",
    "    # End users iteration\n",
    "    \n",
    "    return float('%.3f'%np.mean(store_auc)), float('%.3f'%np.mean(popularity_auc))  \n",
    "   # Return the mean AUC rounded to three decimal places for both test and popularity benchmark"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(0.702, 0.693)"
      ]
     },
     "execution_count": 20,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "calc_mean_auc(product_train, product_users_altered, \n",
    "              [sparse.csr_matrix(user_vecs), sparse.csr_matrix(item_vecs)], product_test)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 推薦系統"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "customers_arr = np.array(customers) \n",
    "products_arr = np.array(products)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "#實際買的\n",
    "def get_items_purchased(customer_id, mf_train, customers_list, products_list, item_lookup):\n",
    "    cust_ind = np.where(customers_list == customer_id)[0][0]\n",
    "    purchased_ind = mf_train[cust_ind,:].nonzero()[1]\n",
    "    prod_codes = products_list[purchased_ind] #得到你買的item id\n",
    "    return item_lookup.loc[item_lookup.item.isin(prod_codes)]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['32009000000620', '32009000001559', '32009000002334',\n",
       "       '32009000006899', '32009000030191'], \n",
       "      dtype='<U14')"
      ]
     },
     "execution_count": 23,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "customers_arr[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 24,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>item</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>8875</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>16450</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>19803</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33</th>\n",
       "      <td>11645</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>47</th>\n",
       "      <td>41855</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>57</th>\n",
       "      <td>19802</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>67</th>\n",
       "      <td>19805</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>70</th>\n",
       "      <td>19941</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>74</th>\n",
       "      <td>8841</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>92</th>\n",
       "      <td>10751</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>130</th>\n",
       "      <td>9465</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>150</th>\n",
       "      <td>16664</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>159</th>\n",
       "      <td>12422</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>186</th>\n",
       "      <td>16663</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>210</th>\n",
       "      <td>11700</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>283</th>\n",
       "      <td>9034</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>349</th>\n",
       "      <td>20701</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>361</th>\n",
       "      <td>15745</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>411</th>\n",
       "      <td>6937</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>427</th>\n",
       "      <td>15547</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>461</th>\n",
       "      <td>8870</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>563</th>\n",
       "      <td>41265</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>583</th>\n",
       "      <td>6095</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>734</th>\n",
       "      <td>19065</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>748</th>\n",
       "      <td>19942</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>757</th>\n",
       "      <td>7650</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>785</th>\n",
       "      <td>9059</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1296</th>\n",
       "      <td>16666</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1434</th>\n",
       "      <td>11148</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1478</th>\n",
       "      <td>19917</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1553</th>\n",
       "      <td>13542</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1564</th>\n",
       "      <td>14408</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1583</th>\n",
       "      <td>42020</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1675</th>\n",
       "      <td>12302</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1731</th>\n",
       "      <td>13963</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1999</th>\n",
       "      <td>9280</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3767</th>\n",
       "      <td>11906</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10060</th>\n",
       "      <td>20662</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>45712</th>\n",
       "      <td>43257</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>121440</th>\n",
       "      <td>45887</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         item\n",
       "4        8875\n",
       "13      16450\n",
       "24      19803\n",
       "33      11645\n",
       "47      41855\n",
       "57      19802\n",
       "67      19805\n",
       "70      19941\n",
       "74       8841\n",
       "92      10751\n",
       "130      9465\n",
       "150     16664\n",
       "159     12422\n",
       "186     16663\n",
       "210     11700\n",
       "283      9034\n",
       "349     20701\n",
       "361     15745\n",
       "411      6937\n",
       "427     15547\n",
       "461      8870\n",
       "563     41265\n",
       "583      6095\n",
       "734     19065\n",
       "748     19942\n",
       "757      7650\n",
       "785      9059\n",
       "1296    16666\n",
       "1434    11148\n",
       "1478    19917\n",
       "1553    13542\n",
       "1564    14408\n",
       "1583    42020\n",
       "1675    12302\n",
       "1731    13963\n",
       "1999     9280\n",
       "3767    11906\n",
       "10060   20662\n",
       "45712   43257\n",
       "121440  45887"
      ]
     },
     "execution_count": 24,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 會員32009000000620所買物品\n",
    "get_items_purchased(\"32009000000620\", product_train, customers_arr, products_arr, item_lookup)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 25,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "from sklearn.preprocessing import MinMaxScaler"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 26,
   "metadata": {
    "collapsed": true
   },
   "outputs": [],
   "source": [
    "# 推薦所買的\n",
    "def rec_items(customer_id, mf_train, user_vecs, item_vecs, customer_list, item_list, item_lookup, num_items = 10):\n",
    "    cust_ind = np.where(customer_list == customer_id)[0][0] # Returns the index row of our customer id\n",
    "    pref_vec = mf_train[cust_ind,:].toarray() # Get the ratings from the training set ratings matrix\n",
    "    pref_vec = pref_vec.reshape(-1) + 1 # Add 1 to everything, so that items not purchased yet become equal to 1\n",
    "    pref_vec[pref_vec > 1] = 0 # Make everything already purchased zero\n",
    "    rec_vector = user_vecs[cust_ind,:].dot(item_vecs.T) # Get dot product of user vector and all item vectors\n",
    "    # Scale this recommendation vector between 0 and 1\n",
    "    min_max = MinMaxScaler()\n",
    "    rec_vector_scaled = min_max.fit_transform(rec_vector.reshape(-1,1))[:,0] \n",
    "    recommend_vector = pref_vec*rec_vector_scaled \n",
    "    # Items already purchased have their recommendation multiplied by zero\n",
    "    product_idx = np.argsort(recommend_vector)[::-1][:num_items] # Sort the indices of the items into order \n",
    "    # of best recommendations\n",
    "    rec_list = [] # start empty list to store items\n",
    "    for index in product_idx:\n",
    "        code = item_list[index]\n",
    "        rec_list.append(code) \n",
    "        # Append our descriptions to the list\n",
    "    codes = [item for item in rec_list]\n",
    "    final_frame = pd.DataFrame({'item': codes}) # Create a dataframe \n",
    "    return final_frame[['item']] # Switch order of columns around"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 27,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>item</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>8838</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>9065</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>9616</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>12163</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>16594</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>19835</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>16792</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>19804</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>8937</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>8842</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "    item\n",
       "0   8838\n",
       "1   9065\n",
       "2   9616\n",
       "3  12163\n",
       "4  16594\n",
       "5  19835\n",
       "6  16792\n",
       "7  19804\n",
       "8   8937\n",
       "9   8842"
      ]
     },
     "execution_count": 27,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# 推薦會員32009000000620前十名\n",
    "rec_items(\"32009000000620\", product_train, user_vecs, item_vecs, customers_arr, products_arr, item_lookup,\n",
    "                       num_items = 10)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 29,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>item</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>9629</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>9063</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>9465</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>9563</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>9065</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   item\n",
       "0  9629\n",
       "1  9063\n",
       "2  9465\n",
       "3  9563\n",
       "4  9065"
      ]
     },
     "execution_count": 29,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rec_items(\"32009000030191\", product_train, user_vecs, item_vecs, customers_arr, products_arr, item_lookup,\n",
    "                       num_items = 5)"
   ]
  }
 ],
 "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": "3.6.1"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}