├── README.md
└── CF Recommendation System - Examples.ipynb


/README.md:
--------------------------------------------------------------------------------
1 | # JNBforBlogs
2 | Jupyter notebooks used in Blogs
3 | 


--------------------------------------------------------------------------------
/CF Recommendation System - Examples.ipynb:
--------------------------------------------------------------------------------
   1 | {
   2 |  "cells": [
   3 |   {
   4 |    "cell_type": "markdown",
   5 |    "metadata": {},
   6 |    "source": [
   7 |     "**Examples of Collaborative Filtering based Recommendation Systems**"
   8 |    ]
   9 |   },
  10 |   {
  11 |    "cell_type": "code",
  12 |    "execution_count": 1,
  13 |    "metadata": {},
  14 |    "outputs": [],
  15 |    "source": [
  16 |     "#make necesarry imports\n",
  17 |     "import numpy as np\n",
  18 |     "import pandas as pd\n",
  19 |     "import matplotlib.pyplot as plt\n",
  20 |     "import sklearn.metrics as metrics\n",
  21 |     "import numpy as np\n",
  22 |     "from sklearn.neighbors import NearestNeighbors\n",
  23 |     "from scipy.spatial.distance import correlation, cosine\n",
  24 |     "import ipywidgets as widgets\n",
  25 |     "from IPython.display import display, clear_output\n",
  26 |     "from sklearn.metrics import pairwise_distances\n",
  27 |     "from sklearn.metrics import mean_squared_error\n",
  28 |     "from math import sqrt\n",
  29 |     "import sys, os\n",
  30 |     "from contextlib import contextmanager"
  31 |    ]
  32 |   },
  33 |   {
  34 |    "cell_type": "code",
  35 |    "execution_count": 2,
  36 |    "metadata": {
  37 |     "collapsed": true
  38 |    },
  39 |    "outputs": [],
  40 |    "source": [
  41 |     "#M is user-item ratings matrix where ratings are integers from 1-10\n",
  42 |     "M = np.asarray([[3,7,4,9,9,7], \n",
  43 |     "                [7,0,5,3,8,8],\n",
  44 |     "               [7,5,5,0,8,4],\n",
  45 |     "               [5,6,8,5,9,8],\n",
  46 |     "               [5,8,8,8,10,9],\n",
  47 |     "               [7,7,0,4,7,8]])\n",
  48 |     "M=pd.DataFrame(M)\n",
  49 |     "\n",
  50 |     "#declaring k,metric as global which can be changed by the user later\n",
  51 |     "global k,metric\n",
  52 |     "k=4\n",
  53 |     "metric='cosine' #can be changed to 'correlation' for Pearson correlation similaries"
  54 |    ]
  55 |   },
  56 |   {
  57 |    "cell_type": "code",
  58 |    "execution_count": 3,
  59 |    "metadata": {},
  60 |    "outputs": [
  61 |     {
  62 |      "data": {
  63 |       "text/html": [
  64 |        "<div>\n",
  65 |        "<style>\n",
  66 |        "    .dataframe thead tr:only-child th {\n",
  67 |        "        text-align: right;\n",
  68 |        "    }\n",
  69 |        "\n",
  70 |        "    .dataframe thead th {\n",
  71 |        "        text-align: left;\n",
  72 |        "    }\n",
  73 |        "\n",
  74 |        "    .dataframe tbody tr th {\n",
  75 |        "        vertical-align: top;\n",
  76 |        "    }\n",
  77 |        "</style>\n",
  78 |        "<table border=\"1\" class=\"dataframe\">\n",
  79 |        "  <thead>\n",
  80 |        "    <tr style=\"text-align: right;\">\n",
  81 |        "      <th></th>\n",
  82 |        "      <th>0</th>\n",
  83 |        "      <th>1</th>\n",
  84 |        "      <th>2</th>\n",
  85 |        "      <th>3</th>\n",
  86 |        "      <th>4</th>\n",
  87 |        "      <th>5</th>\n",
  88 |        "    </tr>\n",
  89 |        "  </thead>\n",
  90 |        "  <tbody>\n",
  91 |        "    <tr>\n",
  92 |        "      <th>0</th>\n",
  93 |        "      <td>3</td>\n",
  94 |        "      <td>7</td>\n",
  95 |        "      <td>4</td>\n",
  96 |        "      <td>9</td>\n",
  97 |        "      <td>9</td>\n",
  98 |        "      <td>7</td>\n",
  99 |        "    </tr>\n",
 100 |        "    <tr>\n",
 101 |        "      <th>1</th>\n",
 102 |        "      <td>7</td>\n",
 103 |        "      <td>0</td>\n",
 104 |        "      <td>5</td>\n",
 105 |        "      <td>3</td>\n",
 106 |        "      <td>8</td>\n",
 107 |        "      <td>8</td>\n",
 108 |        "    </tr>\n",
 109 |        "    <tr>\n",
 110 |        "      <th>2</th>\n",
 111 |        "      <td>7</td>\n",
 112 |        "      <td>5</td>\n",
 113 |        "      <td>5</td>\n",
 114 |        "      <td>0</td>\n",
 115 |        "      <td>8</td>\n",
 116 |        "      <td>4</td>\n",
 117 |        "    </tr>\n",
 118 |        "    <tr>\n",
 119 |        "      <th>3</th>\n",
 120 |        "      <td>5</td>\n",
 121 |        "      <td>6</td>\n",
 122 |        "      <td>8</td>\n",
 123 |        "      <td>5</td>\n",
 124 |        "      <td>9</td>\n",
 125 |        "      <td>8</td>\n",
 126 |        "    </tr>\n",
 127 |        "    <tr>\n",
 128 |        "      <th>4</th>\n",
 129 |        "      <td>5</td>\n",
 130 |        "      <td>8</td>\n",
 131 |        "      <td>8</td>\n",
 132 |        "      <td>8</td>\n",
 133 |        "      <td>10</td>\n",
 134 |        "      <td>9</td>\n",
 135 |        "    </tr>\n",
 136 |        "    <tr>\n",
 137 |        "      <th>5</th>\n",
 138 |        "      <td>7</td>\n",
 139 |        "      <td>7</td>\n",
 140 |        "      <td>0</td>\n",
 141 |        "      <td>4</td>\n",
 142 |        "      <td>7</td>\n",
 143 |        "      <td>8</td>\n",
 144 |        "    </tr>\n",
 145 |        "  </tbody>\n",
 146 |        "</table>\n",
 147 |        "</div>"
 148 |       ],
 149 |       "text/plain": [
 150 |        "   0  1  2  3   4  5\n",
 151 |        "0  3  7  4  9   9  7\n",
 152 |        "1  7  0  5  3   8  8\n",
 153 |        "2  7  5  5  0   8  4\n",
 154 |        "3  5  6  8  5   9  8\n",
 155 |        "4  5  8  8  8  10  9\n",
 156 |        "5  7  7  0  4   7  8"
 157 |       ]
 158 |      },
 159 |      "execution_count": 3,
 160 |      "metadata": {},
 161 |      "output_type": "execute_result"
 162 |     }
 163 |    ],
 164 |    "source": [
 165 |     "M"
 166 |    ]
 167 |   },
 168 |   {
 169 |    "cell_type": "markdown",
 170 |    "metadata": {},
 171 |    "source": [
 172 |     "**User-based Recommendation Systems**"
 173 |    ]
 174 |   },
 175 |   {
 176 |    "cell_type": "code",
 177 |    "execution_count": 4,
 178 |    "metadata": {
 179 |     "collapsed": true
 180 |    },
 181 |    "outputs": [],
 182 |    "source": [
 183 |     "#get cosine similarities for ratings matrix M; pairwise_distances returns the distances between ratings and hence\n",
 184 |     "#similarities are obtained by subtracting distances from 1\n",
 185 |     "cosine_sim = 1-pairwise_distances(M, metric=\"cosine\")"
 186 |    ]
 187 |   },
 188 |   {
 189 |    "cell_type": "code",
 190 |    "execution_count": 5,
 191 |    "metadata": {},
 192 |    "outputs": [
 193 |     {
 194 |      "data": {
 195 |       "text/html": [
 196 |        "<div>\n",
 197 |        "<style>\n",
 198 |        "    .dataframe thead tr:only-child th {\n",
 199 |        "        text-align: right;\n",
 200 |        "    }\n",
 201 |        "\n",
 202 |        "    .dataframe thead th {\n",
 203 |        "        text-align: left;\n",
 204 |        "    }\n",
 205 |        "\n",
 206 |        "    .dataframe tbody tr th {\n",
 207 |        "        vertical-align: top;\n",
 208 |        "    }\n",
 209 |        "</style>\n",
 210 |        "<table border=\"1\" class=\"dataframe\">\n",
 211 |        "  <thead>\n",
 212 |        "    <tr style=\"text-align: right;\">\n",
 213 |        "      <th></th>\n",
 214 |        "      <th>0</th>\n",
 215 |        "      <th>1</th>\n",
 216 |        "      <th>2</th>\n",
 217 |        "      <th>3</th>\n",
 218 |        "      <th>4</th>\n",
 219 |        "      <th>5</th>\n",
 220 |        "    </tr>\n",
 221 |        "  </thead>\n",
 222 |        "  <tbody>\n",
 223 |        "    <tr>\n",
 224 |        "      <th>0</th>\n",
 225 |        "      <td>1.000000</td>\n",
 226 |        "      <td>0.799268</td>\n",
 227 |        "      <td>0.779227</td>\n",
 228 |        "      <td>0.934622</td>\n",
 229 |        "      <td>0.973890</td>\n",
 230 |        "      <td>0.884600</td>\n",
 231 |        "    </tr>\n",
 232 |        "    <tr>\n",
 233 |        "      <th>1</th>\n",
 234 |        "      <td>0.799268</td>\n",
 235 |        "      <td>1.000000</td>\n",
 236 |        "      <td>0.874744</td>\n",
 237 |        "      <td>0.905850</td>\n",
 238 |        "      <td>0.866146</td>\n",
 239 |        "      <td>0.827036</td>\n",
 240 |        "    </tr>\n",
 241 |        "    <tr>\n",
 242 |        "      <th>2</th>\n",
 243 |        "      <td>0.779227</td>\n",
 244 |        "      <td>0.874744</td>\n",
 245 |        "      <td>1.000000</td>\n",
 246 |        "      <td>0.909513</td>\n",
 247 |        "      <td>0.865454</td>\n",
 248 |        "      <td>0.853275</td>\n",
 249 |        "    </tr>\n",
 250 |        "    <tr>\n",
 251 |        "      <th>3</th>\n",
 252 |        "      <td>0.934622</td>\n",
 253 |        "      <td>0.905850</td>\n",
 254 |        "      <td>0.909513</td>\n",
 255 |        "      <td>1.000000</td>\n",
 256 |        "      <td>0.989344</td>\n",
 257 |        "      <td>0.865614</td>\n",
 258 |        "    </tr>\n",
 259 |        "    <tr>\n",
 260 |        "      <th>4</th>\n",
 261 |        "      <td>0.973890</td>\n",
 262 |        "      <td>0.866146</td>\n",
 263 |        "      <td>0.865454</td>\n",
 264 |        "      <td>0.989344</td>\n",
 265 |        "      <td>1.000000</td>\n",
 266 |        "      <td>0.881640</td>\n",
 267 |        "    </tr>\n",
 268 |        "    <tr>\n",
 269 |        "      <th>5</th>\n",
 270 |        "      <td>0.884600</td>\n",
 271 |        "      <td>0.827036</td>\n",
 272 |        "      <td>0.853275</td>\n",
 273 |        "      <td>0.865614</td>\n",
 274 |        "      <td>0.881640</td>\n",
 275 |        "      <td>1.000000</td>\n",
 276 |        "    </tr>\n",
 277 |        "  </tbody>\n",
 278 |        "</table>\n",
 279 |        "</div>"
 280 |       ],
 281 |       "text/plain": [
 282 |        "          0         1         2         3         4         5\n",
 283 |        "0  1.000000  0.799268  0.779227  0.934622  0.973890  0.884600\n",
 284 |        "1  0.799268  1.000000  0.874744  0.905850  0.866146  0.827036\n",
 285 |        "2  0.779227  0.874744  1.000000  0.909513  0.865454  0.853275\n",
 286 |        "3  0.934622  0.905850  0.909513  1.000000  0.989344  0.865614\n",
 287 |        "4  0.973890  0.866146  0.865454  0.989344  1.000000  0.881640\n",
 288 |        "5  0.884600  0.827036  0.853275  0.865614  0.881640  1.000000"
 289 |       ]
 290 |      },
 291 |      "execution_count": 5,
 292 |      "metadata": {},
 293 |      "output_type": "execute_result"
 294 |     }
 295 |    ],
 296 |    "source": [
 297 |     "#Cosine similarity matrix\n",
 298 |     "pd.DataFrame(cosine_sim)"
 299 |    ]
 300 |   },
 301 |   {
 302 |    "cell_type": "code",
 303 |    "execution_count": 6,
 304 |    "metadata": {
 305 |     "collapsed": true
 306 |    },
 307 |    "outputs": [],
 308 |    "source": [
 309 |     "#get pearson similarities for ratings matrix M\n",
 310 |     "pearson_sim = 1-pairwise_distances(M, metric=\"correlation\")"
 311 |    ]
 312 |   },
 313 |   {
 314 |    "cell_type": "code",
 315 |    "execution_count": 7,
 316 |    "metadata": {},
 317 |    "outputs": [
 318 |     {
 319 |      "data": {
 320 |       "text/html": [
 321 |        "<div>\n",
 322 |        "<style>\n",
 323 |        "    .dataframe thead tr:only-child th {\n",
 324 |        "        text-align: right;\n",
 325 |        "    }\n",
 326 |        "\n",
 327 |        "    .dataframe thead th {\n",
 328 |        "        text-align: left;\n",
 329 |        "    }\n",
 330 |        "\n",
 331 |        "    .dataframe tbody tr th {\n",
 332 |        "        vertical-align: top;\n",
 333 |        "    }\n",
 334 |        "</style>\n",
 335 |        "<table border=\"1\" class=\"dataframe\">\n",
 336 |        "  <thead>\n",
 337 |        "    <tr style=\"text-align: right;\">\n",
 338 |        "      <th></th>\n",
 339 |        "      <th>0</th>\n",
 340 |        "      <th>1</th>\n",
 341 |        "      <th>2</th>\n",
 342 |        "      <th>3</th>\n",
 343 |        "      <th>4</th>\n",
 344 |        "      <th>5</th>\n",
 345 |        "    </tr>\n",
 346 |        "  </thead>\n",
 347 |        "  <tbody>\n",
 348 |        "    <tr>\n",
 349 |        "      <th>0</th>\n",
 350 |        "      <td>1.000000</td>\n",
 351 |        "      <td>-0.137446</td>\n",
 352 |        "      <td>-0.357398</td>\n",
 353 |        "      <td>0.208179</td>\n",
 354 |        "      <td>0.761905</td>\n",
 355 |        "      <td>0.277350</td>\n",
 356 |        "    </tr>\n",
 357 |        "    <tr>\n",
 358 |        "      <th>1</th>\n",
 359 |        "      <td>-0.137446</td>\n",
 360 |        "      <td>1.000000</td>\n",
 361 |        "      <td>0.453897</td>\n",
 362 |        "      <td>0.515910</td>\n",
 363 |        "      <td>0.112456</td>\n",
 364 |        "      <td>0.218328</td>\n",
 365 |        "    </tr>\n",
 366 |        "    <tr>\n",
 367 |        "      <th>2</th>\n",
 368 |        "      <td>-0.357398</td>\n",
 369 |        "      <td>0.453897</td>\n",
 370 |        "      <td>1.000000</td>\n",
 371 |        "      <td>0.451378</td>\n",
 372 |        "      <td>-0.042888</td>\n",
 373 |        "      <td>0.297373</td>\n",
 374 |        "    </tr>\n",
 375 |        "    <tr>\n",
 376 |        "      <th>3</th>\n",
 377 |        "      <td>0.208179</td>\n",
 378 |        "      <td>0.515910</td>\n",
 379 |        "      <td>0.451378</td>\n",
 380 |        "      <td>1.000000</td>\n",
 381 |        "      <td>0.763325</td>\n",
 382 |        "      <td>-0.057739</td>\n",
 383 |        "    </tr>\n",
 384 |        "    <tr>\n",
 385 |        "      <th>4</th>\n",
 386 |        "      <td>0.761905</td>\n",
 387 |        "      <td>0.112456</td>\n",
 388 |        "      <td>-0.042888</td>\n",
 389 |        "      <td>0.763325</td>\n",
 390 |        "      <td>1.000000</td>\n",
 391 |        "      <td>0.039621</td>\n",
 392 |        "    </tr>\n",
 393 |        "    <tr>\n",
 394 |        "      <th>5</th>\n",
 395 |        "      <td>0.277350</td>\n",
 396 |        "      <td>0.218328</td>\n",
 397 |        "      <td>0.297373</td>\n",
 398 |        "      <td>-0.057739</td>\n",
 399 |        "      <td>0.039621</td>\n",
 400 |        "      <td>1.000000</td>\n",
 401 |        "    </tr>\n",
 402 |        "  </tbody>\n",
 403 |        "</table>\n",
 404 |        "</div>"
 405 |       ],
 406 |       "text/plain": [
 407 |        "          0         1         2         3         4         5\n",
 408 |        "0  1.000000 -0.137446 -0.357398  0.208179  0.761905  0.277350\n",
 409 |        "1 -0.137446  1.000000  0.453897  0.515910  0.112456  0.218328\n",
 410 |        "2 -0.357398  0.453897  1.000000  0.451378 -0.042888  0.297373\n",
 411 |        "3  0.208179  0.515910  0.451378  1.000000  0.763325 -0.057739\n",
 412 |        "4  0.761905  0.112456 -0.042888  0.763325  1.000000  0.039621\n",
 413 |        "5  0.277350  0.218328  0.297373 -0.057739  0.039621  1.000000"
 414 |       ]
 415 |      },
 416 |      "execution_count": 7,
 417 |      "metadata": {},
 418 |      "output_type": "execute_result"
 419 |     }
 420 |    ],
 421 |    "source": [
 422 |     "#Pearson correlation similarity matrix\n",
 423 |     "pd.DataFrame(pearson_sim)"
 424 |    ]
 425 |   },
 426 |   {
 427 |    "cell_type": "code",
 428 |    "execution_count": 8,
 429 |    "metadata": {
 430 |     "collapsed": true
 431 |    },
 432 |    "outputs": [],
 433 |    "source": [
 434 |     "#This function finds k similar users given the user_id and ratings matrix M\n",
 435 |     "#Note that the similarities are same as obtained via using pairwise_distances\n",
 436 |     "def findksimilarusers(user_id, ratings, metric = metric, k=k):\n",
 437 |     "    similarities=[]\n",
 438 |     "    indices=[]\n",
 439 |     "    model_knn = NearestNeighbors(metric = metric, algorithm = 'brute') \n",
 440 |     "    model_knn.fit(ratings)\n",
 441 |     "\n",
 442 |     "    distances, indices = model_knn.kneighbors(ratings.iloc[user_id-1, :].values.reshape(1, -1), n_neighbors = k)\n",
 443 |     "    similarities = 1-distances.flatten()\n",
 444 |     "    print '{0} most similar users for User {1}:\\n'.format(k-1,user_id)\n",
 445 |     "    for i in range(0, len(indices.flatten())):\n",
 446 |     "        if indices.flatten()[i]+1 == user_id:\n",
 447 |     "            continue;\n",
 448 |     "\n",
 449 |     "        else:\n",
 450 |     "            print '{0}: User {1}, with similarity of {2}'.format(i, indices.flatten()[i]+1, similarities.flatten()[i])\n",
 451 |     "            \n",
 452 |     "    return similarities,indices"
 453 |    ]
 454 |   },
 455 |   {
 456 |    "cell_type": "code",
 457 |    "execution_count": 9,
 458 |    "metadata": {},
 459 |    "outputs": [
 460 |     {
 461 |      "name": "stdout",
 462 |      "output_type": "stream",
 463 |      "text": [
 464 |       "3 most similar users for User 1:\n",
 465 |       "\n",
 466 |       "1: User 5, with similarity of 0.973889935402\n",
 467 |       "2: User 4, with similarity of 0.934621684178\n",
 468 |       "3: User 6, with similarity of 0.88460045723\n"
 469 |      ]
 470 |     }
 471 |    ],
 472 |    "source": [
 473 |     "similarities,indices = findksimilarusers(1,M, metric='cosine')"
 474 |    ]
 475 |   },
 476 |   {
 477 |    "cell_type": "code",
 478 |    "execution_count": 10,
 479 |    "metadata": {
 480 |     "scrolled": true
 481 |    },
 482 |    "outputs": [
 483 |     {
 484 |      "name": "stdout",
 485 |      "output_type": "stream",
 486 |      "text": [
 487 |       "3 most similar users for User 1:\n",
 488 |       "\n",
 489 |       "1: User 5, with similarity of 0.761904761905\n",
 490 |       "2: User 6, with similarity of 0.277350098113\n",
 491 |       "3: User 4, with similarity of 0.208179450927\n"
 492 |      ]
 493 |     }
 494 |    ],
 495 |    "source": [
 496 |     "similarities,indices = findksimilarusers(1,M, metric='correlation')"
 497 |    ]
 498 |   },
 499 |   {
 500 |    "cell_type": "code",
 501 |    "execution_count": 11,
 502 |    "metadata": {
 503 |     "collapsed": true
 504 |    },
 505 |    "outputs": [],
 506 |    "source": [
 507 |     "#This function predicts rating for specified user-item combination based on user-based approach\n",
 508 |     "def predict_userbased(user_id, item_id, ratings, metric = metric, k=k):\n",
 509 |     "    prediction=0\n",
 510 |     "    similarities, indices=findksimilarusers(user_id, ratings,metric, k) #similar users based on cosine similarity\n",
 511 |     "    mean_rating = ratings.loc[user_id-1,:].mean() #to adjust for zero based indexing\n",
 512 |     "    sum_wt = np.sum(similarities)-1\n",
 513 |     "    product=1\n",
 514 |     "    wtd_sum = 0 \n",
 515 |     "    \n",
 516 |     "    for i in range(0, len(indices.flatten())):\n",
 517 |     "        if indices.flatten()[i]+1 == user_id:\n",
 518 |     "            continue;\n",
 519 |     "        else: \n",
 520 |     "            ratings_diff = ratings.iloc[indices.flatten()[i],item_id-1]-np.mean(ratings.iloc[indices.flatten()[i],:])\n",
 521 |     "            product = ratings_diff * (similarities[i])\n",
 522 |     "            wtd_sum = wtd_sum + product\n",
 523 |     "    \n",
 524 |     "    prediction = int(round(mean_rating + (wtd_sum/sum_wt)))\n",
 525 |     "    print '\\nPredicted rating for user {0} -> item {1}: {2}'.format(user_id,item_id,prediction)\n",
 526 |     "\n",
 527 |     "    return prediction"
 528 |    ]
 529 |   },
 530 |   {
 531 |    "cell_type": "code",
 532 |    "execution_count": 12,
 533 |    "metadata": {},
 534 |    "outputs": [
 535 |     {
 536 |      "name": "stdout",
 537 |      "output_type": "stream",
 538 |      "text": [
 539 |       "3 most similar users for User 3:\n",
 540 |       "\n",
 541 |       "1: User 4, with similarity of 0.90951268934\n",
 542 |       "2: User 2, with similarity of 0.874744414849\n",
 543 |       "3: User 5, with similarity of 0.86545387815\n",
 544 |       "\n",
 545 |       "Predicted rating for user 3 -> item 4: 3\n"
 546 |      ]
 547 |     }
 548 |    ],
 549 |    "source": [
 550 |     "predict_userbased(3,4,M);"
 551 |    ]
 552 |   },
 553 |   {
 554 |    "cell_type": "markdown",
 555 |    "metadata": {},
 556 |    "source": [
 557 |     "**Item-based Recommendation Systems**"
 558 |    ]
 559 |   },
 560 |   {
 561 |    "cell_type": "code",
 562 |    "execution_count": 13,
 563 |    "metadata": {
 564 |     "collapsed": true
 565 |    },
 566 |    "outputs": [],
 567 |    "source": [
 568 |     "#This function finds k similar items given the item_id and ratings matrix M\n",
 569 |     "\n",
 570 |     "def findksimilaritems(item_id, ratings, metric=metric, k=k):\n",
 571 |     "    similarities=[]\n",
 572 |     "    indices=[]    \n",
 573 |     "    ratings=ratings.T\n",
 574 |     "    model_knn = NearestNeighbors(metric = metric, algorithm = 'brute')\n",
 575 |     "    model_knn.fit(ratings)\n",
 576 |     "\n",
 577 |     "    distances, indices = model_knn.kneighbors(ratings.iloc[item_id-1, :].values.reshape(1, -1), n_neighbors = k)\n",
 578 |     "    similarities = 1-distances.flatten()\n",
 579 |     "    print '{0} most similar items for item {1}:\\n'.format(k-1,item_id)\n",
 580 |     "    for i in range(0, len(indices.flatten())):\n",
 581 |     "        if indices.flatten()[i]+1 == item_id:\n",
 582 |     "            continue;\n",
 583 |     "\n",
 584 |     "        else:\n",
 585 |     "            print '{0}: Item {1} :, with similarity of {2}'.format(i,indices.flatten()[i]+1, similarities.flatten()[i])\n",
 586 |     "\n",
 587 |     "\n",
 588 |     "    return similarities,indices"
 589 |    ]
 590 |   },
 591 |   {
 592 |    "cell_type": "code",
 593 |    "execution_count": 14,
 594 |    "metadata": {},
 595 |    "outputs": [
 596 |     {
 597 |      "name": "stdout",
 598 |      "output_type": "stream",
 599 |      "text": [
 600 |       "3 most similar items for item 3:\n",
 601 |       "\n",
 602 |       "1: Item 5 :, with similarity of 0.918336125535\n",
 603 |       "2: Item 6 :, with similarity of 0.874759773038\n",
 604 |       "3: Item 1 :, with similarity of 0.810364746222\n"
 605 |      ]
 606 |     }
 607 |    ],
 608 |    "source": [
 609 |     "similarities,indices=findksimilaritems(3,M)"
 610 |    ]
 611 |   },
 612 |   {
 613 |    "cell_type": "code",
 614 |    "execution_count": 15,
 615 |    "metadata": {
 616 |     "collapsed": true
 617 |    },
 618 |    "outputs": [],
 619 |    "source": [
 620 |     "#This function predicts the rating for specified user-item combination based on item-based approach\n",
 621 |     "def predict_itembased(user_id, item_id, ratings, metric = metric, k=k):\n",
 622 |     "    prediction= wtd_sum =0\n",
 623 |     "    similarities, indices=findksimilaritems(item_id, ratings) #similar users based on correlation coefficients\n",
 624 |     "    sum_wt = np.sum(similarities)-1\n",
 625 |     "    product=1\n",
 626 |     "    \n",
 627 |     "    for i in range(0, len(indices.flatten())):\n",
 628 |     "        if indices.flatten()[i]+1 == item_id:\n",
 629 |     "            continue;\n",
 630 |     "        else:\n",
 631 |     "            product = ratings.iloc[user_id-1,indices.flatten()[i]] * (similarities[i])\n",
 632 |     "            wtd_sum = wtd_sum + product                              \n",
 633 |     "    prediction = int(round(wtd_sum/sum_wt))\n",
 634 |     "    print '\\nPredicted rating for user {0} -> item {1}: {2}'.format(user_id,item_id,prediction)      \n",
 635 |     "\n",
 636 |     "    return prediction"
 637 |    ]
 638 |   },
 639 |   {
 640 |    "cell_type": "code",
 641 |    "execution_count": 16,
 642 |    "metadata": {},
 643 |    "outputs": [
 644 |     {
 645 |      "name": "stdout",
 646 |      "output_type": "stream",
 647 |      "text": [
 648 |       "3 most similar items for item 3:\n",
 649 |       "\n",
 650 |       "1: Item 5 :, with similarity of 0.918336125535\n",
 651 |       "2: Item 6 :, with similarity of 0.874759773038\n",
 652 |       "3: Item 1 :, with similarity of 0.810364746222\n",
 653 |       "\n",
 654 |       "Predicted rating for user 1 -> item 3: 6\n"
 655 |      ]
 656 |     }
 657 |    ],
 658 |    "source": [
 659 |     "prediction = predict_itembased(1,3,M)"
 660 |    ]
 661 |   },
 662 |   {
 663 |    "cell_type": "code",
 664 |    "execution_count": 17,
 665 |    "metadata": {},
 666 |    "outputs": [],
 667 |    "source": [
 668 |     "#This function is used to compute adjusted cosine similarity matrix for items\n",
 669 |     "def computeAdjCosSim(M):\n",
 670 |     "    sim_matrix = np.zeros((M.shape[1], M.shape[1]))\n",
 671 |     "    M_u = M.mean(axis=1) #means\n",
 672 |     "          \n",
 673 |     "    for i in range(M.shape[1]):\n",
 674 |     "        for j in range(M.shape[1]):\n",
 675 |     "            if i == j:\n",
 676 |     "                \n",
 677 |     "                sim_matrix[i][j] = 1\n",
 678 |     "            else:                \n",
 679 |     "                if i<j:\n",
 680 |     "                    \n",
 681 |     "                    sum_num = sum_den1 = sum_den2 = 0\n",
 682 |     "                    for k,row in M.loc[:,[i,j]].iterrows(): \n",
 683 |     "\n",
 684 |     "                        if ((M.loc[k,i] != 0) & (M.loc[k,j] != 0)):\n",
 685 |     "                            num = (M[i][k]-M_u[k])*(M[j][k]-M_u[k])\n",
 686 |     "                            den1= (M[i][k]-M_u[k])**2\n",
 687 |     "                            den2= (M[j][k]-M_u[k])**2\n",
 688 |     "                            \n",
 689 |     "                            sum_num = sum_num + num\n",
 690 |     "                            sum_den1 = sum_den1 + den1\n",
 691 |     "                            sum_den2 = sum_den2 + den2\n",
 692 |     "                        \n",
 693 |     "                        else:\n",
 694 |     "                            continue                          \n",
 695 |     "                                       \n",
 696 |     "                    den=(sum_den1**0.5)*(sum_den2**0.5)\n",
 697 |     "                    if den!=0:\n",
 698 |     "                        sim_matrix[i][j] = sum_num/den\n",
 699 |     "                    else:\n",
 700 |     "                        sim_matrix[i][j] = 0\n",
 701 |     "\n",
 702 |     "\n",
 703 |     "                else:\n",
 704 |     "                    sim_matrix[i][j] = sim_matrix[j][i]           \n",
 705 |     "            \n",
 706 |     "    return pd.DataFrame(sim_matrix)"
 707 |    ]
 708 |   },
 709 |   {
 710 |    "cell_type": "code",
 711 |    "execution_count": 18,
 712 |    "metadata": {},
 713 |    "outputs": [],
 714 |    "source": [
 715 |     "adjcos_sim = computeAdjCosSim(M)"
 716 |    ]
 717 |   },
 718 |   {
 719 |    "cell_type": "code",
 720 |    "execution_count": 19,
 721 |    "metadata": {},
 722 |    "outputs": [
 723 |     {
 724 |      "data": {
 725 |       "text/html": [
 726 |        "<div>\n",
 727 |        "<style>\n",
 728 |        "    .dataframe thead tr:only-child th {\n",
 729 |        "        text-align: right;\n",
 730 |        "    }\n",
 731 |        "\n",
 732 |        "    .dataframe thead th {\n",
 733 |        "        text-align: left;\n",
 734 |        "    }\n",
 735 |        "\n",
 736 |        "    .dataframe tbody tr th {\n",
 737 |        "        vertical-align: top;\n",
 738 |        "    }\n",
 739 |        "</style>\n",
 740 |        "<table border=\"1\" class=\"dataframe\">\n",
 741 |        "  <thead>\n",
 742 |        "    <tr style=\"text-align: right;\">\n",
 743 |        "      <th></th>\n",
 744 |        "      <th>0</th>\n",
 745 |        "      <th>1</th>\n",
 746 |        "      <th>2</th>\n",
 747 |        "      <th>3</th>\n",
 748 |        "      <th>4</th>\n",
 749 |        "      <th>5</th>\n",
 750 |        "    </tr>\n",
 751 |        "  </thead>\n",
 752 |        "  <tbody>\n",
 753 |        "    <tr>\n",
 754 |        "      <th>0</th>\n",
 755 |        "      <td>1.000000</td>\n",
 756 |        "      <td>0.236908</td>\n",
 757 |        "      <td>0.421263</td>\n",
 758 |        "      <td>-0.519085</td>\n",
 759 |        "      <td>-0.125892</td>\n",
 760 |        "      <td>0.010090</td>\n",
 761 |        "    </tr>\n",
 762 |        "    <tr>\n",
 763 |        "      <th>1</th>\n",
 764 |        "      <td>0.236908</td>\n",
 765 |        "      <td>1.000000</td>\n",
 766 |        "      <td>-0.805243</td>\n",
 767 |        "      <td>0.085741</td>\n",
 768 |        "      <td>0.237273</td>\n",
 769 |        "      <td>0.520625</td>\n",
 770 |        "    </tr>\n",
 771 |        "    <tr>\n",
 772 |        "      <th>2</th>\n",
 773 |        "      <td>0.421263</td>\n",
 774 |        "      <td>-0.805243</td>\n",
 775 |        "      <td>1.000000</td>\n",
 776 |        "      <td>-0.767941</td>\n",
 777 |        "      <td>-0.230521</td>\n",
 778 |        "      <td>-0.053640</td>\n",
 779 |        "    </tr>\n",
 780 |        "    <tr>\n",
 781 |        "      <th>3</th>\n",
 782 |        "      <td>-0.519085</td>\n",
 783 |        "      <td>0.085741</td>\n",
 784 |        "      <td>-0.767941</td>\n",
 785 |        "      <td>1.000000</td>\n",
 786 |        "      <td>-0.299059</td>\n",
 787 |        "      <td>-0.644550</td>\n",
 788 |        "    </tr>\n",
 789 |        "    <tr>\n",
 790 |        "      <th>4</th>\n",
 791 |        "      <td>-0.125892</td>\n",
 792 |        "      <td>0.237273</td>\n",
 793 |        "      <td>-0.230521</td>\n",
 794 |        "      <td>-0.299059</td>\n",
 795 |        "      <td>1.000000</td>\n",
 796 |        "      <td>0.599158</td>\n",
 797 |        "    </tr>\n",
 798 |        "    <tr>\n",
 799 |        "      <th>5</th>\n",
 800 |        "      <td>0.010090</td>\n",
 801 |        "      <td>0.520625</td>\n",
 802 |        "      <td>-0.053640</td>\n",
 803 |        "      <td>-0.644550</td>\n",
 804 |        "      <td>0.599158</td>\n",
 805 |        "      <td>1.000000</td>\n",
 806 |        "    </tr>\n",
 807 |        "  </tbody>\n",
 808 |        "</table>\n",
 809 |        "</div>"
 810 |       ],
 811 |       "text/plain": [
 812 |        "          0         1         2         3         4         5\n",
 813 |        "0  1.000000  0.236908  0.421263 -0.519085 -0.125892  0.010090\n",
 814 |        "1  0.236908  1.000000 -0.805243  0.085741  0.237273  0.520625\n",
 815 |        "2  0.421263 -0.805243  1.000000 -0.767941 -0.230521 -0.053640\n",
 816 |        "3 -0.519085  0.085741 -0.767941  1.000000 -0.299059 -0.644550\n",
 817 |        "4 -0.125892  0.237273 -0.230521 -0.299059  1.000000  0.599158\n",
 818 |        "5  0.010090  0.520625 -0.053640 -0.644550  0.599158  1.000000"
 819 |       ]
 820 |      },
 821 |      "execution_count": 19,
 822 |      "metadata": {},
 823 |      "output_type": "execute_result"
 824 |     }
 825 |    ],
 826 |    "source": [
 827 |     "adjcos_sim"
 828 |    ]
 829 |   },
 830 |   {
 831 |    "cell_type": "code",
 832 |    "execution_count": 20,
 833 |    "metadata": {
 834 |     "collapsed": true
 835 |    },
 836 |    "outputs": [],
 837 |    "source": [
 838 |     "#This function finds k similar items given the item_id and ratings matrix M\n",
 839 |     "\n",
 840 |     "def findksimilaritems_adjcos(item_id, ratings, k=k):\n",
 841 |     "    \n",
 842 |     "    sim_matrix = computeAdjCosSim(ratings)\n",
 843 |     "    similarities = sim_matrix[item_id-1].sort_values(ascending=False)[:k].values\n",
 844 |     "    indices = sim_matrix[item_id-1].sort_values(ascending=False)[:k].index\n",
 845 |     "    \n",
 846 |     "    print '{0} most similar items for item {1}:\\n'.format(k-1,item_id)\n",
 847 |     "    for i in range(0, len(indices)):\n",
 848 |     "            if indices[i]+1 == item_id:\n",
 849 |     "                continue;\n",
 850 |     "\n",
 851 |     "            else:\n",
 852 |     "                print '{0}: Item {1} :, with similarity of {2}'.format(i,indices[i]+1, similarities[i])\n",
 853 |     "        \n",
 854 |     "    return similarities ,indices    "
 855 |    ]
 856 |   },
 857 |   {
 858 |    "cell_type": "code",
 859 |    "execution_count": 21,
 860 |    "metadata": {},
 861 |    "outputs": [
 862 |     {
 863 |      "name": "stdout",
 864 |      "output_type": "stream",
 865 |      "text": [
 866 |       "3 most similar items for item 3:\n",
 867 |       "\n",
 868 |       "1: Item 1 :, with similarity of 0.421262731871\n",
 869 |       "2: Item 6 :, with similarity of -0.0536398904889\n",
 870 |       "3: Item 5 :, with similarity of -0.230521358269\n"
 871 |      ]
 872 |     }
 873 |    ],
 874 |    "source": [
 875 |     "similarities, indices = findksimilaritems_adjcos(3,M)"
 876 |    ]
 877 |   },
 878 |   {
 879 |    "cell_type": "code",
 880 |    "execution_count": 22,
 881 |    "metadata": {
 882 |     "collapsed": true
 883 |    },
 884 |    "outputs": [],
 885 |    "source": [
 886 |     "#This function predicts the rating for specified user-item combination for adjusted cosine item-based approach\n",
 887 |     "#As the adjusted cosine similarities range from -1,+1, sometimes the predicted rating can be negative or greater than max value\n",
 888 |     "#Hack to deal with this: Rating is set to min if prediction is negative, Rating is set to max if prediction is above max\n",
 889 |     "def predict_itembased_adjcos(user_id, item_id, ratings):\n",
 890 |     "    prediction=0\n",
 891 |     "\n",
 892 |     "    similarities, indices=findksimilaritems_adjcos(item_id, ratings) #similar users based on correlation coefficients\n",
 893 |     "    sum_wt = np.sum(similarities)-1\n",
 894 |     "\n",
 895 |     "    product=1\n",
 896 |     "    wtd_sum = 0 \n",
 897 |     "    for i in range(0, len(indices)):\n",
 898 |     "        if indices[i]+1 == item_id:\n",
 899 |     "            continue;\n",
 900 |     "        else:\n",
 901 |     "            product = ratings.iloc[user_id-1,indices[i]] * (similarities[i])\n",
 902 |     "            wtd_sum = wtd_sum + product                              \n",
 903 |     "    prediction = int(round(wtd_sum/sum_wt))\n",
 904 |     "    if prediction < 0:\n",
 905 |     "        prediction = 1\n",
 906 |     "    elif prediction >10:\n",
 907 |     "        prediction = 10\n",
 908 |     "    print '\\nPredicted rating for user {0} -> item {1}: {2}'.format(user_id,item_id,prediction)      \n",
 909 |     "        \n",
 910 |     "    return prediction"
 911 |    ]
 912 |   },
 913 |   {
 914 |    "cell_type": "code",
 915 |    "execution_count": 23,
 916 |    "metadata": {},
 917 |    "outputs": [
 918 |     {
 919 |      "name": "stdout",
 920 |      "output_type": "stream",
 921 |      "text": [
 922 |       "3 most similar items for item 4:\n",
 923 |       "\n",
 924 |       "1: Item 2 :, with similarity of 0.0857414341149\n",
 925 |       "2: Item 5 :, with similarity of -0.29905882779\n",
 926 |       "3: Item 1 :, with similarity of -0.519085268895\n",
 927 |       "\n",
 928 |       "Predicted rating for user 3 -> item 4: 8\n"
 929 |      ]
 930 |     }
 931 |    ],
 932 |    "source": [
 933 |     "prediction=predict_itembased_adjcos(3,4,M)"
 934 |    ]
 935 |   },
 936 |   {
 937 |    "cell_type": "code",
 938 |    "execution_count": 24,
 939 |    "metadata": {},
 940 |    "outputs": [
 941 |     {
 942 |      "data": {
 943 |       "text/html": [
 944 |        "<div>\n",
 945 |        "<style>\n",
 946 |        "    .dataframe thead tr:only-child th {\n",
 947 |        "        text-align: right;\n",
 948 |        "    }\n",
 949 |        "\n",
 950 |        "    .dataframe thead th {\n",
 951 |        "        text-align: left;\n",
 952 |        "    }\n",
 953 |        "\n",
 954 |        "    .dataframe tbody tr th {\n",
 955 |        "        vertical-align: top;\n",
 956 |        "    }\n",
 957 |        "</style>\n",
 958 |        "<table border=\"1\" class=\"dataframe\">\n",
 959 |        "  <thead>\n",
 960 |        "    <tr style=\"text-align: right;\">\n",
 961 |        "      <th></th>\n",
 962 |        "      <th>0</th>\n",
 963 |        "      <th>1</th>\n",
 964 |        "      <th>2</th>\n",
 965 |        "      <th>3</th>\n",
 966 |        "      <th>4</th>\n",
 967 |        "      <th>5</th>\n",
 968 |        "    </tr>\n",
 969 |        "  </thead>\n",
 970 |        "  <tbody>\n",
 971 |        "    <tr>\n",
 972 |        "      <th>0</th>\n",
 973 |        "      <td>1.000000</td>\n",
 974 |        "      <td>0.236908</td>\n",
 975 |        "      <td>0.421263</td>\n",
 976 |        "      <td>-0.519085</td>\n",
 977 |        "      <td>-0.125892</td>\n",
 978 |        "      <td>0.010090</td>\n",
 979 |        "    </tr>\n",
 980 |        "    <tr>\n",
 981 |        "      <th>1</th>\n",
 982 |        "      <td>0.236908</td>\n",
 983 |        "      <td>1.000000</td>\n",
 984 |        "      <td>-0.805243</td>\n",
 985 |        "      <td>0.085741</td>\n",
 986 |        "      <td>0.237273</td>\n",
 987 |        "      <td>0.520625</td>\n",
 988 |        "    </tr>\n",
 989 |        "    <tr>\n",
 990 |        "      <th>2</th>\n",
 991 |        "      <td>0.421263</td>\n",
 992 |        "      <td>-0.805243</td>\n",
 993 |        "      <td>1.000000</td>\n",
 994 |        "      <td>-0.767941</td>\n",
 995 |        "      <td>-0.230521</td>\n",
 996 |        "      <td>-0.053640</td>\n",
 997 |        "    </tr>\n",
 998 |        "    <tr>\n",
 999 |        "      <th>3</th>\n",
1000 |        "      <td>-0.519085</td>\n",
1001 |        "      <td>0.085741</td>\n",
1002 |        "      <td>-0.767941</td>\n",
1003 |        "      <td>1.000000</td>\n",
1004 |        "      <td>-0.299059</td>\n",
1005 |        "      <td>-0.644550</td>\n",
1006 |        "    </tr>\n",
1007 |        "    <tr>\n",
1008 |        "      <th>4</th>\n",
1009 |        "      <td>-0.125892</td>\n",
1010 |        "      <td>0.237273</td>\n",
1011 |        "      <td>-0.230521</td>\n",
1012 |        "      <td>-0.299059</td>\n",
1013 |        "      <td>1.000000</td>\n",
1014 |        "      <td>0.599158</td>\n",
1015 |        "    </tr>\n",
1016 |        "    <tr>\n",
1017 |        "      <th>5</th>\n",
1018 |        "      <td>0.010090</td>\n",
1019 |        "      <td>0.520625</td>\n",
1020 |        "      <td>-0.053640</td>\n",
1021 |        "      <td>-0.644550</td>\n",
1022 |        "      <td>0.599158</td>\n",
1023 |        "      <td>1.000000</td>\n",
1024 |        "    </tr>\n",
1025 |        "  </tbody>\n",
1026 |        "</table>\n",
1027 |        "</div>"
1028 |       ],
1029 |       "text/plain": [
1030 |        "          0         1         2         3         4         5\n",
1031 |        "0  1.000000  0.236908  0.421263 -0.519085 -0.125892  0.010090\n",
1032 |        "1  0.236908  1.000000 -0.805243  0.085741  0.237273  0.520625\n",
1033 |        "2  0.421263 -0.805243  1.000000 -0.767941 -0.230521 -0.053640\n",
1034 |        "3 -0.519085  0.085741 -0.767941  1.000000 -0.299059 -0.644550\n",
1035 |        "4 -0.125892  0.237273 -0.230521 -0.299059  1.000000  0.599158\n",
1036 |        "5  0.010090  0.520625 -0.053640 -0.644550  0.599158  1.000000"
1037 |       ]
1038 |      },
1039 |      "execution_count": 24,
1040 |      "metadata": {},
1041 |      "output_type": "execute_result"
1042 |     }
1043 |    ],
1044 |    "source": [
1045 |     "adjcos_sim"
1046 |    ]
1047 |   },
1048 |   {
1049 |    "cell_type": "code",
1050 |    "execution_count": 25,
1051 |    "metadata": {
1052 |     "collapsed": true
1053 |    },
1054 |    "outputs": [],
1055 |    "source": [
1056 |     "#This function utilizes above function to recommend items for selected approach. Recommendations are made if the predicted\n",
1057 |     "#rating for an item is greater than or equal to 6, and the items has not been rated already\n",
1058 |     "def recommendItem(user_id, item_id, ratings):\n",
1059 |     "    \n",
1060 |     "    if user_id<1 or user_id>6 or type(user_id) is not int:\n",
1061 |     "        print 'Userid does not exist. Enter numbers from 1-6'\n",
1062 |     "    else:    \n",
1063 |     "        ids = ['User-based CF (cosine)','User-based CF (correlation)','Item-based CF (cosine)',\n",
1064 |     "               'Item-based CF (adjusted cosine)']\n",
1065 |     "\n",
1066 |     "        approach = widgets.Dropdown(options=ids, value=ids[0],\n",
1067 |     "                               description='Select Approach', width='500px')\n",
1068 |     "        \n",
1069 |     "        def on_change(change):\n",
1070 |     "            prediction = 0\n",
1071 |     "            clear_output(wait=True)\n",
1072 |     "            if change['type'] == 'change' and change['name'] == 'value':            \n",
1073 |     "                if (approach.value == 'User-based CF (cosine)'):\n",
1074 |     "                    metric = 'cosine'\n",
1075 |     "                    prediction = predict_userbased(user_id, item_id, ratings, metric)\n",
1076 |     "                elif (approach.value == 'User-based CF (correlation)')  :                       \n",
1077 |     "                    metric = 'correlation'               \n",
1078 |     "                    prediction = predict_userbased(user_id, item_id, ratings, metric)\n",
1079 |     "                elif (approach.value == 'Item-based CF (cosine)'):\n",
1080 |     "                    prediction = predict_itembased(user_id, item_id, ratings)\n",
1081 |     "                else:\n",
1082 |     "                    prediction = predict_itembased_adjcos(user_id,item_id,ratings)\n",
1083 |     "\n",
1084 |     "                if ratings[item_id-1][user_id-1] != 0: \n",
1085 |     "                    print 'Item already rated'\n",
1086 |     "                else:\n",
1087 |     "                    if prediction>=6:\n",
1088 |     "                        print '\\nItem recommended'\n",
1089 |     "                    else:\n",
1090 |     "                        print 'Item not recommended'\n",
1091 |     "\n",
1092 |     "        approach.observe(on_change)\n",
1093 |     "        display(approach)"
1094 |    ]
1095 |   },
1096 |   {
1097 |    "cell_type": "code",
1098 |    "execution_count": 26,
1099 |    "metadata": {},
1100 |    "outputs": [
1101 |     {
1102 |      "name": "stdout",
1103 |      "output_type": "stream",
1104 |      "text": [
1105 |       "Userid does not exist. Enter numbers from 1-6\n"
1106 |      ]
1107 |     }
1108 |    ],
1109 |    "source": [
1110 |     "#check for incorrect entries\n",
1111 |     "recommendItem(-1,3,M)"
1112 |    ]
1113 |   },
1114 |   {
1115 |    "cell_type": "code",
1116 |    "execution_count": 27,
1117 |    "metadata": {},
1118 |    "outputs": [
1119 |     {
1120 |      "name": "stdout",
1121 |      "output_type": "stream",
1122 |      "text": [
1123 |       "3 most similar users for User 3:\n",
1124 |       "\n",
1125 |       "1: User 4, with similarity of 0.90951268934\n",
1126 |       "2: User 2, with similarity of 0.874744414849\n",
1127 |       "3: User 5, with similarity of 0.86545387815\n",
1128 |       "\n",
1129 |       "Predicted rating for user 3 -> item 4: 3\n",
1130 |       "Item not recommended\n"
1131 |      ]
1132 |     }
1133 |    ],
1134 |    "source": [
1135 |     "recommendItem(3,4,M)"
1136 |    ]
1137 |   },
1138 |   {
1139 |    "cell_type": "code",
1140 |    "execution_count": 28,
1141 |    "metadata": {},
1142 |    "outputs": [
1143 |     {
1144 |      "name": "stdout",
1145 |      "output_type": "stream",
1146 |      "text": [
1147 |       "3 most similar users for User 3:\n",
1148 |       "\n",
1149 |       "1: User 2, with similarity of 0.453897185842\n",
1150 |       "2: User 4, with similarity of 0.451378005098\n",
1151 |       "3: User 6, with similarity of 0.297373304825\n",
1152 |       "\n",
1153 |       "Predicted rating for user 3 -> item 4: 3\n",
1154 |       "Item not recommended\n"
1155 |      ]
1156 |     }
1157 |    ],
1158 |    "source": [
1159 |     "recommendItem(3,4,M)"
1160 |    ]
1161 |   },
1162 |   {
1163 |    "cell_type": "code",
1164 |    "execution_count": 29,
1165 |    "metadata": {},
1166 |    "outputs": [
1167 |     {
1168 |      "name": "stdout",
1169 |      "output_type": "stream",
1170 |      "text": [
1171 |       "3 most similar items for item 4:\n",
1172 |       "\n",
1173 |       "1: Item 6 :, with similarity of 0.89977997614\n",
1174 |       "2: Item 2 :, with similarity of 0.887160079571\n",
1175 |       "3: Item 5 :, with similarity of 0.88180009273\n",
1176 |       "\n",
1177 |       "Predicted rating for user 3 -> item 4: 6\n",
1178 |       "\n",
1179 |       "Item recommended\n"
1180 |      ]
1181 |     }
1182 |    ],
1183 |    "source": [
1184 |     "recommendItem(3,4,M)"
1185 |    ]
1186 |   },
1187 |   {
1188 |    "cell_type": "code",
1189 |    "execution_count": 30,
1190 |    "metadata": {},
1191 |    "outputs": [
1192 |     {
1193 |      "name": "stdout",
1194 |      "output_type": "stream",
1195 |      "text": [
1196 |       "3 most similar items for item 4:\n",
1197 |       "\n",
1198 |       "1: Item 2 :, with similarity of 0.0857414341149\n",
1199 |       "2: Item 5 :, with similarity of -0.29905882779\n",
1200 |       "3: Item 1 :, with similarity of -0.519085268895\n",
1201 |       "\n",
1202 |       "Predicted rating for user 3 -> item 4: 8\n",
1203 |       "\n",
1204 |       "Item recommended\n"
1205 |      ]
1206 |     }
1207 |    ],
1208 |    "source": [
1209 |     "recommendItem(3,4,M)"
1210 |    ]
1211 |   },
1212 |   {
1213 |    "cell_type": "code",
1214 |    "execution_count": 31,
1215 |    "metadata": {},
1216 |    "outputs": [
1217 |     {
1218 |      "name": "stdout",
1219 |      "output_type": "stream",
1220 |      "text": [
1221 |       "3 most similar users for User 2:\n",
1222 |       "\n",
1223 |       "1: User 4, with similarity of 0.515910067398\n",
1224 |       "2: User 3, with similarity of 0.453897185842\n",
1225 |       "3: User 6, with similarity of 0.218327934565\n",
1226 |       "\n",
1227 |       "Predicted rating for user 2 -> item 1: 5\n",
1228 |       "Item already rated\n"
1229 |      ]
1230 |     }
1231 |    ],
1232 |    "source": [
1233 |     "#if the item is already rated, it is not recommended\n",
1234 |     "recommendItem(2,1,M)"
1235 |    ]
1236 |   },
1237 |   {
1238 |    "cell_type": "code",
1239 |    "execution_count": 35,
1240 |    "metadata": {},
1241 |    "outputs": [],
1242 |    "source": [
1243 |     "#This is a quick way to temporarily suppress stdout in particular code section\n",
1244 |     "@contextmanager\n",
1245 |     "def suppress_stdout():\n",
1246 |     "    with open(os.devnull, \"w\") as devnull:\n",
1247 |     "        old_stdout = sys.stdout\n",
1248 |     "        sys.stdout = devnull\n",
1249 |     "        try:  \n",
1250 |     "            yield\n",
1251 |     "        finally:\n",
1252 |     "            sys.stdout = old_stdout"
1253 |    ]
1254 |   },
1255 |   {
1256 |    "cell_type": "code",
1257 |    "execution_count": 45,
1258 |    "metadata": {},
1259 |    "outputs": [],
1260 |    "source": [
1261 |     "#This is final function to evaluate the performance of selected recommendation approach and the metric used here is RMSE\n",
1262 |     "#suppress_stdout function is used to suppress the print outputs of all the functions inside this function. It will only print \n",
1263 |     "#RMSE values\n",
1264 |     "def evaluateRS(ratings):\n",
1265 |     "    ids = ['User-based CF (cosine)','User-based CF (correlation)','Item-based CF (cosine)','Item-based CF (adjusted cosine)']\n",
1266 |     "    approach = widgets.Dropdown(options=ids, value=ids[0],description='Select Approach', width='500px')\n",
1267 |     "    n_users = ratings.shape[0]\n",
1268 |     "    n_items = ratings.shape[1]\n",
1269 |     "    prediction = np.zeros((n_users, n_items))\n",
1270 |     "    prediction= pd.DataFrame(prediction)\n",
1271 |     "    def on_change(change):\n",
1272 |     "        clear_output(wait=True)\n",
1273 |     "        with suppress_stdout():\n",
1274 |     "            if change['type'] == 'change' and change['name'] == 'value':            \n",
1275 |     "                if (approach.value == 'User-based CF (cosine)'):\n",
1276 |     "                    metric = 'cosine'\n",
1277 |     "                    for i in range(n_users):\n",
1278 |     "                        for j in range(n_items):\n",
1279 |     "                            prediction[i][j] = predict_userbased(i+1, j+1, ratings, metric)\n",
1280 |     "                elif (approach.value == 'User-based CF (correlation)')  :                       \n",
1281 |     "                    metric = 'correlation'               \n",
1282 |     "                    for i in range(n_users):\n",
1283 |     "                        for j in range(n_items):\n",
1284 |     "                            prediction[i][j] = predict_userbased(i+1, j+1, ratings, metric)\n",
1285 |     "                elif (approach.value == 'Item-based CF (cosine)'):\n",
1286 |     "                    for i in range(n_users):\n",
1287 |     "                        for j in range(n_items):\n",
1288 |     "                            prediction[i][j] = predict_userbased(i+1, j+1, ratings)\n",
1289 |     "                else:\n",
1290 |     "                    for i in range(n_users):\n",
1291 |     "                        for j in range(n_items):\n",
1292 |     "                            prediction[i][j] = predict_userbased(i+1, j+1, ratings)\n",
1293 |     "              \n",
1294 |     "        MSE = mean_squared_error(prediction, ratings)\n",
1295 |     "        RMSE = round(sqrt(MSE),3)\n",
1296 |     "        print \"RMSE using {0} approach is: {1}\".format(approach.value,RMSE)\n",
1297 |     "              \n",
1298 |     "    approach.observe(on_change)\n",
1299 |     "    display(approach)"
1300 |    ]
1301 |   },
1302 |   {
1303 |    "cell_type": "code",
1304 |    "execution_count": 46,
1305 |    "metadata": {},
1306 |    "outputs": [
1307 |     {
1308 |      "name": "stdout",
1309 |      "output_type": "stream",
1310 |      "text": [
1311 |       "RMSE using User-based CF (cosine) approach is: 2.667\n"
1312 |      ]
1313 |     }
1314 |    ],
1315 |    "source": [
1316 |     "evaluateRS(M)"
1317 |    ]
1318 |   },
1319 |   {
1320 |    "cell_type": "code",
1321 |    "execution_count": 47,
1322 |    "metadata": {},
1323 |    "outputs": [
1324 |     {
1325 |      "name": "stdout",
1326 |      "output_type": "stream",
1327 |      "text": [
1328 |       "RMSE using User-based CF (correlation) approach is: 2.764\n"
1329 |      ]
1330 |     }
1331 |    ],
1332 |    "source": [
1333 |     "evaluateRS(M)"
1334 |    ]
1335 |   },
1336 |   {
1337 |    "cell_type": "markdown",
1338 |    "metadata": {},
1339 |    "source": [
1340 |     "**Thanks for reading this notebook**"
1341 |    ]
1342 |   }
1343 |  ],
1344 |  "metadata": {
1345 |   "kernelspec": {
1346 |    "display_name": "Python 2",
1347 |    "language": "python",
1348 |    "name": "python2"
1349 |   },
1350 |   "language_info": {
1351 |    "codemirror_mode": {
1352 |     "name": "ipython",
1353 |     "version": 2
1354 |    },
1355 |    "file_extension": ".py",
1356 |    "mimetype": "text/x-python",
1357 |    "name": "python",
1358 |    "nbconvert_exporter": "python",
1359 |    "pygments_lexer": "ipython2",
1360 |    "version": "2.7.13"
1361 |   }
1362 |  },
1363 |  "nbformat": 4,
1364 |  "nbformat_minor": 2
1365 | }
1366 | 


--------------------------------------------------------------------------------