Secure group recommendations WIP.ipynb

{
 "cells": [
  {
   "cell_type": "markdown",
   "id": "5baa75d7",
   "metadata": {},
   "source": [
    "Mean of $N$ variables $\\cal{N}(0, \\sigma)$ is $\\cal{N}(0, \\sigma/\\sqrt{n})$"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 1,
   "id": "ff0bf104",
   "metadata": {},
   "outputs": [],
   "source": [
    "from mangaki.models import Profile"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "id": "a8bae07e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Counter({False: 92, True: 7950})"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from collections import Counter\n",
    "\n",
    "Counter(Profile.objects.values_list('is_shared', flat=True))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "id": "33a5cde1",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.011439940313354886"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "92/(7950+92)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "id": "20df525f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Counter({(False, True): 78,\n",
       "         (True, True): 7833,\n",
       "         (False, False): 14,\n",
       "         (True, False): 117})"
      ]
     },
     "execution_count": 13,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Counter(list(Profile.objects.values_list('is_shared', 'research_ok')))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 368,
   "id": "3834bc43",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<QuerySet [{'user': 372}]>"
      ]
     },
     "execution_count": 368,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "Profile.objects.filter(is_shared=False, research_ok=True).order_by('?').values('user')[:1]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 433,
   "id": "447a5c90",
   "metadata": {},
   "outputs": [],
   "source": [
    "PRIVATE_USER = 372"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 434,
   "id": "f30d9aa0",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_private = pd.DataFrame(\n",
    "    User.objects.get(id=372).rating_set.values_list('work_id', 'choice'),\n",
    "    columns=('item_id', 'choice')).sort_values('item_id')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 435,
   "id": "a07468e6",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "417"
      ]
     },
     "execution_count": 435,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(df_private)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 436,
   "id": "d96a532d",
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "\n",
    "df = pd.DataFrame(Rating.objects.exclude(user_id=PRIVATE_USER).filter(\n",
    "    # user__profile__is_shared=False,\n",
    "    user__profile__research_ok=True,\n",
    "    work__in=df_private['item_id']).values_list(\n",
    "    'user_id', 'work_id', 'choice'), columns=('user_id', 'item_id', 'choice'))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 437,
   "id": "ce964294",
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(   user_id  item_id   choice\n",
       " 0      169        1     like\n",
       " 1      229        1     like\n",
       " 2      224        1     like\n",
       " 3      278        1  willsee\n",
       " 4      234        1  dislike,\n",
       " 61684)"
      ]
     },
     "execution_count": 437,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head(), len(df)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 444,
   "id": "b9cebc73",
   "metadata": {},
   "outputs": [],
   "source": [
    "item_ids, cols = np.unique(df['item_id'], return_inverse=True)\n",
    "encode_col = dict(zip(df['item_id'], cols))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 445,
   "id": "e22572d2",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(417, 61684)"
      ]
     },
     "execution_count": 445,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(item_ids), len(cols)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 447,
   "id": "345d496a",
   "metadata": {},
   "outputs": [],
   "source": [
    "from mangaki.utils.values import rating_values\n",
    "import numpy as np\n",
    "\n",
    "df['row'] = np.unique(df['user_id'], return_inverse=True)[1]\n",
    "df['col'] = np.unique(df['item_id'], return_inverse=True)[1]\n",
    "df['value'] = df['choice'].map(rating_values)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 448,
   "id": "4b36ea48",
   "metadata": {},
   "outputs": [],
   "source": [
    "df_private['value'] = df_private['choice'].map(rating_values)\n",
    "df_private['col'] = df_private['item_id'].map(encode_col)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 449,
   "id": "994d4830",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.model_selection import train_test_split"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 450,
   "id": "a06faf7d",
   "metadata": {},
   "outputs": [],
   "source": [
    "private_train, private_test = train_test_split(df_private, test_size=0.2, shuffle=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 452,
   "id": "deee1e8a",
   "metadata": {},
   "outputs": [],
   "source": [
    "private_train = private_train.sort_values('col')\n",
    "private_test = private_test.sort_values('col')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 453,
   "id": "ff3c7015",
   "metadata": {},
   "outputs": [],
   "source": [
    "train = np.array(private_train['col'])\n",
    "test = np.array(private_test['col'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 454,
   "id": "57790a50",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>user_id</th>\n",
       "      <th>item_id</th>\n",
       "      <th>choice</th>\n",
       "      <th>row</th>\n",
       "      <th>col</th>\n",
       "      <th>value</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>169</td>\n",
       "      <td>1</td>\n",
       "      <td>like</td>\n",
       "      <td>139</td>\n",
       "      <td>0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>229</td>\n",
       "      <td>1</td>\n",
       "      <td>like</td>\n",
       "      <td>189</td>\n",
       "      <td>0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>224</td>\n",
       "      <td>1</td>\n",
       "      <td>like</td>\n",
       "      <td>184</td>\n",
       "      <td>0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>278</td>\n",
       "      <td>1</td>\n",
       "      <td>willsee</td>\n",
       "      <td>231</td>\n",
       "      <td>0</td>\n",
       "      <td>0.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>234</td>\n",
       "      <td>1</td>\n",
       "      <td>dislike</td>\n",
       "      <td>194</td>\n",
       "      <td>0</td>\n",
       "      <td>-2.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   user_id  item_id   choice  row  col  value\n",
       "0      169        1     like  139    0    2.0\n",
       "1      229        1     like  189    0    2.0\n",
       "2      224        1     like  184    0    2.0\n",
       "3      278        1  willsee  231    0    0.5\n",
       "4      234        1  dislike  194    0   -2.0"
      ]
     },
     "execution_count": 454,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 455,
   "id": "ba7dc50a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>item_id</th>\n",
       "      <th>col</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1751</th>\n",
       "      <td>2</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2984</th>\n",
       "      <td>4</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4450</th>\n",
       "      <td>5</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6074</th>\n",
       "      <td>6</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>61157</th>\n",
       "      <td>8012</td>\n",
       "      <td>412</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>61459</th>\n",
       "      <td>8019</td>\n",
       "      <td>413</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>61466</th>\n",
       "      <td>8050</td>\n",
       "      <td>414</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>61572</th>\n",
       "      <td>8051</td>\n",
       "      <td>415</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>61619</th>\n",
       "      <td>8062</td>\n",
       "      <td>416</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>417 rows × 2 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       item_id  col\n",
       "0            1    0\n",
       "1751         2    1\n",
       "2984         4    2\n",
       "4450         5    3\n",
       "6074         6    4\n",
       "...        ...  ...\n",
       "61157     8012  412\n",
       "61459     8019  413\n",
       "61466     8050  414\n",
       "61572     8051  415\n",
       "61619     8062  416\n",
       "\n",
       "[417 rows x 2 columns]"
      ]
     },
     "execution_count": 455,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df[['item_id', 'col']].drop_duplicates().sort_values('item_id')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 456,
   "id": "9121615f",
   "metadata": {},
   "outputs": [],
   "source": [
    "from scipy.sparse import csr_matrix\n",
    "\n",
    "M = csr_matrix((df['value'], (df['row'], df['col'])))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 457,
   "id": "20b7017a",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<2156x417 sparse matrix of type '<class 'numpy.float64'>'\n",
       "\twith 61684 stored elements in Compressed Sparse Row format>"
      ]
     },
     "execution_count": 457,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "M"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 458,
   "id": "97450f39",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.linear_model import LinearRegression"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 459,
   "id": "46f451c8",
   "metadata": {},
   "outputs": [],
   "source": [
    "truth = np.array(df_private['value'])\n",
    "train_truth = np.array(private_train['value'])\n",
    "test_truth = np.array(private_test['value'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 460,
   "id": "d3b13d4f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "LinearRegression()"
      ]
     },
     "execution_count": 460,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lr = LinearRegression()\n",
    "lr.fit(M.T[train], train_truth)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 461,
   "id": "d747bab1",
   "metadata": {},
   "outputs": [],
   "source": [
    "lr2 = LinearRegression()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 462,
   "id": "45087dbd",
   "metadata": {},
   "outputs": [],
   "source": [
    "lr3 = LinearRegression()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 463,
   "id": "1cc098ae",
   "metadata": {},
   "outputs": [],
   "source": [
    "pred = lr.predict(M.T)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 464,
   "id": "a35f1feb",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.02075742, 0.02075742, 0.02075742, 0.02075742, 0.02075742,\n",
       "       0.02075742, 0.02075742, 0.02075742, 0.02075742, 0.02075742])"
      ]
     },
     "execution_count": 464,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "weights[:10]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 465,
   "id": "45fea0f3",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "{'favorite': 4,\n",
       " 'like': 2,\n",
       " 'dislike': -2,\n",
       " 'neutral': 0.1,\n",
       " 'willsee': 0.5,\n",
       " 'wontsee': -0.5}"
      ]
     },
     "execution_count": 465,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rating_values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 467,
   "id": "b6943e48",
   "metadata": {},
   "outputs": [],
   "source": [
    "def loss_f(true, pred):\n",
    "    pred = np.clip(pred, -2, 4)\n",
    "    return ((np.round(pred, 3) - true) ** 2).mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 476,
   "id": "1bb0795d",
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "CPU times: user 174 ms, sys: 5.43 ms, total: 179 ms\n",
      "Wall time: 178 ms\n"
     ]
    }
   ],
   "source": [
    "%%time\n",
    "y = []\n",
    "x = [5, 10, 15, 20, 30, 50, 100]\n",
    "y_knn = []  # Train losses\n",
    "y_lr_knn = []\n",
    "y_lr = []\n",
    "y_w_knn = []\n",
    "test_untrained_lr = []\n",
    "test_knn = []  # Test losses\n",
    "test_lr = []\n",
    "test_lr_knn = []\n",
    "test_w_knn = []\n",
    "\n",
    "for KEEP in x:\n",
    "    truncated = np.abs(lr.coef_).argsort()[-KEEP:]\n",
    "    train_approx = M.T[train][:, truncated] @ lr.coef_[truncated] + lr.intercept_\n",
    "    test_approx = M.T[test][:, truncated] @ lr.coef_[truncated] + lr.intercept_\n",
    "    train_loss = loss_f(train_truth, train_approx)\n",
    "    test_loss = loss_f(test_truth, test_approx)\n",
    "    y.append(train_loss)\n",
    "    test_untrained_lr.append(test_loss)\n",
    "    \n",
    "    lr3.fit(M.T[train][:, truncated], train_truth)\n",
    "    train_approx = lr3.predict(M.T[train][:, truncated])\n",
    "    test_approx = lr3.predict(M.T[test][:, truncated])\n",
    "    train_loss = loss_f(train_truth, train_approx)\n",
    "    test_loss = loss_f(test_truth, test_approx)\n",
    "    y_lr.append(train_loss)\n",
    "    test_lr.append(test_loss)\n",
    "    \n",
    "    truncated = neighbors[:KEEP].reshape(-1)\n",
    "    lr2.fit(M.T[train][:, truncated], train_truth)\n",
    "    train_approx = lr2.predict(M.T[train][:, truncated])\n",
    "    test_approx = lr2.predict(M.T[test][:, truncated])\n",
    "    train_loss = loss_f(train_truth, train_approx)\n",
    "    test_loss = loss_f(test_truth, test_approx)\n",
    "    y_lr_knn.append(train_loss)\n",
    "    test_lr_knn.append(test_loss)\n",
    "    \n",
    "    train_approx = M.T[train][:, truncated].mean(axis=1).A1.reshape(-1)\n",
    "    test_approx = M.T[test][:, truncated].mean(axis=1).A1.reshape(-1)\n",
    "    train_loss = loss_f(train_truth, train_approx)\n",
    "    test_loss = loss_f(test_truth, test_approx)\n",
    "    y_knn.append(train_loss)\n",
    "    test_knn.append(test_loss)\n",
    "    \n",
    "    weights = 1 - dist[:KEEP]\n",
    "    train_approx = (M.T[train][:, truncated] * diags(weights)).sum(axis=1).A1.reshape(-1) / weights.sum()\n",
    "    test_approx = (M.T[test][:, truncated] * diags(weights)).sum(axis=1).A1.reshape(-1) / weights.sum()\n",
    "    train_loss = loss_f(train_truth, train_approx)\n",
    "    test_loss = loss_f(test_truth, test_approx)\n",
    "    y_w_knn.append(train_loss)\n",
    "    test_w_knn.append(test_loss)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 477,
   "id": "e055344e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'Train error for one private user')"
      ]
     },
     "execution_count": 477,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#plt.plot(x, y, label='LR not retrained')\n",
    "plt.plot(x, y_knn, label='knn')\n",
    "plt.plot(x, y_w_knn, label='weighted knn')\n",
    "plt.plot(x, y_lr_knn, label='LR knn')\n",
    "plt.plot(x, y_lr, label='LR retrained')\n",
    "#plt.xlim(xmax=500)\n",
    "#plt.ylim(ymax=5)\n",
    "plt.legend()\n",
    "plt.xlabel('# users (neighbors)')\n",
    "plt.ylabel('Train MSE')\n",
    "plt.title('Train error for one private user')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 478,
   "id": "62368251",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0.5, 1.0, 'Test error for one private user')"
      ]
     },
     "execution_count": 478,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "#plt.plot(x, test_untrained_lr, label='LR not retrained')\n",
    "plt.plot(x, test_knn, label='knn')\n",
    "plt.plot(x, test_w_knn, label='weighted knn')\n",
    "plt.plot(x, test_lr_knn, label='LR knn')\n",
    "plt.plot(x, test_lr, label='LR retrained')\n",
    "#plt.xlim(xmax=500)\n",
    "#plt.ylim(ymax=5)\n",
    "plt.legend()\n",
    "plt.xlabel('# users (neighbors)')\n",
    "plt.ylabel('Test MSE')\n",
    "plt.title('Test error for one private user')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 151,
   "id": "f6b9129b",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(array([   6.,   18.,   48.,  257., 1521.,  178.,   36.,   16.,    6.,\n",
       "           6.]),\n",
       " array([-0.10496153, -0.0811011 , -0.05724068, -0.03338026, -0.00951983,\n",
       "         0.01434059,  0.03820101,  0.06206144,  0.08592186,  0.10978228,\n",
       "         0.13364271]),\n",
       " <BarContainer object of 10 artists>)"
      ]
     },
     "execution_count": 151,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "import matplotlib.pyplot as plt\n",
    "\n",
    "plt.hist(lr.coef_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 128,
   "id": "bc62b0d9",
   "metadata": {},
   "outputs": [],
   "source": [
    "best_pos = lr.coef_.argsort()[-5:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 129,
   "id": "e3680606",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([  39,   76,  180, 1176,  475])"
      ]
     },
     "execution_count": 129,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "best_pos"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 130,
   "id": "366a9080",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.11645822, 0.11968056, 0.1199534 , 0.12968687, 0.13364271])"
      ]
     },
     "execution_count": 130,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lr.coef_[best_pos]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 134,
   "id": "dac469c9",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>user_id</th>\n",
       "      <th>item_id</th>\n",
       "      <th>choice</th>\n",
       "      <th>row</th>\n",
       "      <th>col</th>\n",
       "      <th>value</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>9362</th>\n",
       "      <td>557</td>\n",
       "      <td>7</td>\n",
       "      <td>like</td>\n",
       "      <td>475</td>\n",
       "      <td>0</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9363</th>\n",
       "      <td>557</td>\n",
       "      <td>80</td>\n",
       "      <td>dislike</td>\n",
       "      <td>475</td>\n",
       "      <td>13</td>\n",
       "      <td>-2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9364</th>\n",
       "      <td>557</td>\n",
       "      <td>742</td>\n",
       "      <td>willsee</td>\n",
       "      <td>475</td>\n",
       "      <td>38</td>\n",
       "      <td>0.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9365</th>\n",
       "      <td>557</td>\n",
       "      <td>954</td>\n",
       "      <td>like</td>\n",
       "      <td>475</td>\n",
       "      <td>44</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9366</th>\n",
       "      <td>557</td>\n",
       "      <td>1377</td>\n",
       "      <td>wontsee</td>\n",
       "      <td>475</td>\n",
       "      <td>52</td>\n",
       "      <td>-0.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>...</th>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "      <td>...</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>36730</th>\n",
       "      <td>557</td>\n",
       "      <td>2040</td>\n",
       "      <td>neutral</td>\n",
       "      <td>475</td>\n",
       "      <td>71</td>\n",
       "      <td>0.1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>36898</th>\n",
       "      <td>557</td>\n",
       "      <td>2078</td>\n",
       "      <td>wontsee</td>\n",
       "      <td>475</td>\n",
       "      <td>75</td>\n",
       "      <td>-0.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37111</th>\n",
       "      <td>557</td>\n",
       "      <td>410</td>\n",
       "      <td>like</td>\n",
       "      <td>475</td>\n",
       "      <td>31</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>40196</th>\n",
       "      <td>557</td>\n",
       "      <td>151</td>\n",
       "      <td>willsee</td>\n",
       "      <td>475</td>\n",
       "      <td>23</td>\n",
       "      <td>0.5</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>40329</th>\n",
       "      <td>557</td>\n",
       "      <td>1289</td>\n",
       "      <td>like</td>\n",
       "      <td>475</td>\n",
       "      <td>48</td>\n",
       "      <td>2.0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>84 rows × 6 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "       user_id  item_id   choice  row  col  value\n",
       "9362       557        7     like  475    0    2.0\n",
       "9363       557       80  dislike  475   13   -2.0\n",
       "9364       557      742  willsee  475   38    0.5\n",
       "9365       557      954     like  475   44    2.0\n",
       "9366       557     1377  wontsee  475   52   -0.5\n",
       "...        ...      ...      ...  ...  ...    ...\n",
       "36730      557     2040  neutral  475   71    0.1\n",
       "36898      557     2078  wontsee  475   75   -0.5\n",
       "37111      557      410     like  475   31    2.0\n",
       "40196      557      151  willsee  475   23    0.5\n",
       "40329      557     1289     like  475   48    2.0\n",
       "\n",
       "[84 rows x 6 columns]"
      ]
     },
     "execution_count": 134,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df.query(\"row == 475\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 175,
   "id": "1e32f399",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics.pairwise import cosine_similarity\n",
    "\n",
    "sim = cosine_similarity(np.array(df_private['value']).reshape(1, -1), M)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 176,
   "id": "52904469",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([[ 106,  241, 1208,   76,  343]])"
      ]
     },
     "execution_count": 176,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sim.argsort()[:, -5:]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 177,
   "id": "5ea1d45e",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.3728813559322034"
      ]
     },
     "execution_count": 177,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(M[343].toarray()[0] == truth).mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 178,
   "id": "76e4ec09",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.4491525423728814"
      ]
     },
     "execution_count": 178,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(M[343].toarray()[0] == 0).mean()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 468,
   "id": "eab82ccb",
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.neighbors import KNeighborsRegressor, NearestNeighbors\n",
    "\n",
    "knn = NearestNeighbors(n_neighbors=1000, metric='cosine')\n",
    "knn.fit(M[:, train])\n",
    "\n",
    "dist, neighbors = knn.kneighbors(train_truth.reshape(1, -1))\n",
    "dist = dist[0]\n",
    "neighbors = neighbors[0]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 410,
   "id": "75a149ee",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([1434, 1555,  117, 1562,  114])"
      ]
     },
     "execution_count": 410,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "neighbors[:5]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 411,
   "id": "31d72d86",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([0.86179229])"
      ]
     },
     "execution_count": 411,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "1 - M[343] @ truth / (np.linalg.norm(truth) * np.linalg.norm(M[343].toarray()))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 232,
   "id": "7194775f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "[<matplotlib.lines.Line2D at 0x7fc48f12cbb0>]"
      ]
     },
     "execution_count": 232,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    }
   ],
   "source": [
    "plt.plot(dist[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 236,
   "id": "b7df1c7f",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(1, 1000)"
      ]
     },
     "execution_count": 236,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "neighbors.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 479,
   "id": "83b4c027",
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "1.61245154965971"
      ]
     },
     "execution_count": 479,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "2.6 ** 0.5"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "id": "6a263a1b",
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Django Shell-Plus",
   "language": "python",
   "name": "django_extensions"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.9.5"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 5
}