rank.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [],
   "source": [
    "import carball\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "import os\n",
    "import glob\n",
    "import torch\n",
    "import torch.nn as nn\n",
    "import torch.optim as optim\n",
    "import requests\n",
    "%matplotlib inline"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 408,
   "metadata": {},
   "outputs": [],
   "source": [
    "def get_playergames(rank=19):\n",
    "    url = \"https://calculated.gg/api/v1/playergames?key=1&rank={}&playlist=13\".format(rank)\n",
    "    r = requests.get(url)\n",
    "    return r.json()\n",
    "\n",
    "def get_playergames_by_id(id_=\"76561198018756583\", all_=False, days=30):\n",
    "    url = \"https://calculated.gg/api/v1/playergames?key=1&player_ids={}&playlist=13&days={}\".format(id_, days)\n",
    "    r = requests.get(url)\n",
    "    print(url)\n",
    "    js = r.json()\n",
    "    if not all_:\n",
    "        js['data'] = [d for d in js['data'] if d[2] == id_]\n",
    "    return js\n",
    "\n",
    "smurfs = ['76561198142512330', '76561198867867878', '76561198039062964']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 602,
   "metadata": {},
   "outputs": [],
   "source": [
    "def concat_data(rank_datas):\n",
    "    input_data = None\n",
    "    for rank_data in rank_datas:\n",
    "        data = pd.DataFrame(rank_data['data'], columns=rank_data['columns']).fillna(0)\n",
    "        if input_data is not None:\n",
    "            input_data = input_data.append(data)\n",
    "        else:\n",
    "            input_data = data\n",
    "    return input_data.reset_index()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 241,
   "metadata": {},
   "outputs": [],
   "source": [
    "datas = []\n",
    "for rank in range(10, 20):\n",
    "    for x in range(5):\n",
    "        rank_data = get_playergames(rank)\n",
    "        datas.append(rank_data)\n",
    "            \n",
    "input_data = concat_data(datas)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 603,
   "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>index</th>\n",
       "      <th>id</th>\n",
       "      <th>name</th>\n",
       "      <th>player</th>\n",
       "      <th>game</th>\n",
       "      <th>rank</th>\n",
       "      <th>division</th>\n",
       "      <th>mmr</th>\n",
       "      <th>is_orange</th>\n",
       "      <th>win</th>\n",
       "      <th>...</th>\n",
       "      <th>time_closest_to_team_center</th>\n",
       "      <th>time_furthest_from_team_center</th>\n",
       "      <th>is_bot</th>\n",
       "      <th>first_frame_in_game</th>\n",
       "      <th>time_in_game</th>\n",
       "      <th>time_in_front_of_center_of_mass</th>\n",
       "      <th>time_behind_center_of_mass</th>\n",
       "      <th>time_most_forward_player</th>\n",
       "      <th>time_most_back_player</th>\n",
       "      <th>time_between_players</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>1511218</td>\n",
       "      <td>dunkphat</td>\n",
       "      <td>76561197968853721</td>\n",
       "      <td>A7A2B24C11E8CDEAF36DB4B695A970F4</td>\n",
       "      <td>10</td>\n",
       "      <td>2</td>\n",
       "      <td>748</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>False</td>\n",
       "      <td>1.0</td>\n",
       "      <td>311.794128</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>1846453</td>\n",
       "      <td>dunkphat</td>\n",
       "      <td>76561197968853721</td>\n",
       "      <td>DF1857B411E8D104D40F24971E6114F5</td>\n",
       "      <td>10</td>\n",
       "      <td>3</td>\n",
       "      <td>758</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>False</td>\n",
       "      <td>1.0</td>\n",
       "      <td>320.648499</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>2358199</td>\n",
       "      <td>felagund9</td>\n",
       "      <td>76561198104052806</td>\n",
       "      <td>7E75A22411E8DA352CC87B812D90E774</td>\n",
       "      <td>10</td>\n",
       "      <td>1</td>\n",
       "      <td>640</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>58.87915</td>\n",
       "      <td>39.764526</td>\n",
       "      <td>False</td>\n",
       "      <td>1.0</td>\n",
       "      <td>322.324982</td>\n",
       "      <td>145.612656</td>\n",
       "      <td>176.712341</td>\n",
       "      <td>105.340714</td>\n",
       "      <td>127.388824</td>\n",
       "      <td>89.595444</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>1265012</td>\n",
       "      <td>Floboss le Francais</td>\n",
       "      <td>76561197991597902</td>\n",
       "      <td>AF6C467949A99892969057B3C25493DD</td>\n",
       "      <td>10</td>\n",
       "      <td>0</td>\n",
       "      <td>637</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>False</td>\n",
       "      <td>1.0</td>\n",
       "      <td>364.692932</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>1029522</td>\n",
       "      <td>Callanmari</td>\n",
       "      <td>76561198169600688</td>\n",
       "      <td>06BC313943F0863B6A17C3A68D3B78E5</td>\n",
       "      <td>10</td>\n",
       "      <td>3</td>\n",
       "      <td>760</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>0.00000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>False</td>\n",
       "      <td>1.0</td>\n",
       "      <td>336.128082</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>5 rows × 82 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "   index       id                 name             player  \\\n",
       "0      0  1511218             dunkphat  76561197968853721   \n",
       "1      1  1846453             dunkphat  76561197968853721   \n",
       "2      2  2358199            felagund9  76561198104052806   \n",
       "3      3  1265012  Floboss le Francais  76561197991597902   \n",
       "4      4  1029522           Callanmari  76561198169600688   \n",
       "\n",
       "                               game  rank  division  mmr  is_orange    win  \\\n",
       "0  A7A2B24C11E8CDEAF36DB4B695A970F4    10         2  748       True   True   \n",
       "1  DF1857B411E8D104D40F24971E6114F5    10         3  758      False  False   \n",
       "2  7E75A22411E8DA352CC87B812D90E774    10         1  640      False  False   \n",
       "3  AF6C467949A99892969057B3C25493DD    10         0  637      False  False   \n",
       "4  06BC313943F0863B6A17C3A68D3B78E5    10         3  760      False   True   \n",
       "\n",
       "           ...           time_closest_to_team_center  \\\n",
       "0          ...                               0.00000   \n",
       "1          ...                               0.00000   \n",
       "2          ...                              58.87915   \n",
       "3          ...                               0.00000   \n",
       "4          ...                               0.00000   \n",
       "\n",
       "   time_furthest_from_team_center  is_bot  first_frame_in_game  time_in_game  \\\n",
       "0                        0.000000   False                  1.0    311.794128   \n",
       "1                        0.000000   False                  1.0    320.648499   \n",
       "2                       39.764526   False                  1.0    322.324982   \n",
       "3                        0.000000   False                  1.0    364.692932   \n",
       "4                        0.000000   False                  1.0    336.128082   \n",
       "\n",
       "   time_in_front_of_center_of_mass  time_behind_center_of_mass  \\\n",
       "0                         0.000000                    0.000000   \n",
       "1                         0.000000                    0.000000   \n",
       "2                       145.612656                  176.712341   \n",
       "3                         0.000000                    0.000000   \n",
       "4                         0.000000                    0.000000   \n",
       "\n",
       "   time_most_forward_player  time_most_back_player  time_between_players  \n",
       "0                  0.000000               0.000000              0.000000  \n",
       "1                  0.000000               0.000000              0.000000  \n",
       "2                105.340714             127.388824             89.595444  \n",
       "3                  0.000000               0.000000              0.000000  \n",
       "4                  0.000000               0.000000              0.000000  \n",
       "\n",
       "[5 rows x 82 columns]"
      ]
     },
     "execution_count": 603,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "\n",
    "\n",
    "input_data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 616,
   "metadata": {},
   "outputs": [],
   "source": [
    "\n",
    "def process_input_data(input_, maxs=None, mins=None):\n",
    "    input_columns = ['saves', 'total_aerials', \n",
    "                     'time_at_boost_speed', 'time_at_slow_speed', 'average_speed', \n",
    "                     'boost_usage', 'average_hit_distance', 'ball_hit_forward', 'won_turnovers',\n",
    "                     'num_stolen_boosts']#\n",
    "    ignore = ['total_saves', 'is_bot', 'time_in_game', 'car']\n",
    "    input_columns = list(input_.columns[11:])\n",
    "    nonzero = input_[input_['time_in_game'] > 0]\n",
    "    input = nonzero[input_columns].div(nonzero['time_in_game'], axis=0)\n",
    "    input = input.drop(ignore, axis=1)\n",
    "#     for c in input.columns:\n",
    "#         if (input[c].max()-input[c].min()) == 0:\n",
    "#             print(c)\n",
    "#             input = input.drop(c, axis=1)\n",
    "    if maxs is None or mins is None:\n",
    "        mins = input.min()\n",
    "        maxs = input.max()\n",
    "    input = (input-mins)/(maxs - mins)\n",
    "    output = nonzero['rank'].values.reshape((-1, 1)).astype(float)\n",
    "    return input, output, maxs, mins\n",
    "input, output, maxs, mins = process_input_data(input_data)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 617,
   "metadata": {},
   "outputs": [],
   "source": [
    "threshold = np.random.rand(len(output)) < 0.9\n",
    "input_train, input_test = input[threshold].values.astype(float), input[~threshold].values.astype(float)\n",
    "output_train, output_test = output[threshold], output[~threshold]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 342,
   "metadata": {},
   "outputs": [],
   "source": [
    "num_columns = len(input.columns)\n",
    "n_hidden = 128\n",
    "class RankPredictor(nn.Module):\n",
    "    def __init__(self):\n",
    "        super().__init__()\n",
    "        self.layers = nn.Sequential(\n",
    "            nn.Linear(num_columns, n_hidden),\n",
    "            nn.ReLU(),\n",
    "            nn.Linear(n_hidden, n_hidden//2),\n",
    "            nn.ReLU(),\n",
    "            nn.Linear(n_hidden//2, n_hidden//4),\n",
    "            nn.ReLU(),\n",
    "            nn.Linear(n_hidden//4, 1)\n",
    "        )\n",
    "    \n",
    "    def forward(self, game_state):\n",
    "        game_state = torch.from_numpy(game_state).float().cuda()\n",
    "        return self.layers(game_state)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 343,
   "metadata": {},
   "outputs": [
    {
     "ename": "KeyboardInterrupt",
     "evalue": "",
     "output_type": "error",
     "traceback": [
      "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m                         Traceback (most recent call last)",
      "\u001b[0;32m<ipython-input-343-6f9a4b15fc00>\u001b[0m in \u001b[0;36m<module>\u001b[0;34m\u001b[0m\n\u001b[1;32m     11\u001b[0m \u001b[0;32mfor\u001b[0m \u001b[0mepoch\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mrange\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mepochs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     12\u001b[0m     \u001b[0mopt\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mzero_grad\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 13\u001b[0;31m     \u001b[0mpredicted\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mmodel\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0minput_train\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     14\u001b[0m     \u001b[0mloss_val\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mloss\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mpredicted\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0moutput_tensor\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     15\u001b[0m     \u001b[0mloss_val\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mbackward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m~/PycharmProjects/ReplayML/venv/lib/python3.6/site-packages/torch/nn/modules/module.py\u001b[0m in \u001b[0;36m__call__\u001b[0;34m(self, *input, **kwargs)\u001b[0m\n\u001b[1;32m    475\u001b[0m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_slow_forward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    476\u001b[0m         \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 477\u001b[0;31m             \u001b[0mresult\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mforward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m*\u001b[0m\u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0;34m**\u001b[0m\u001b[0mkwargs\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m    478\u001b[0m         \u001b[0;32mfor\u001b[0m \u001b[0mhook\u001b[0m \u001b[0;32min\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_forward_hooks\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mvalues\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m    479\u001b[0m             \u001b[0mhook_result\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mhook\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0minput\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mresult\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;32m<ipython-input-342-05597b7b08f7>\u001b[0m in \u001b[0;36mforward\u001b[0;34m(self, game_state)\u001b[0m\n\u001b[1;32m     15\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m     16\u001b[0m     \u001b[0;32mdef\u001b[0m \u001b[0mforward\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mself\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mgame_state\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m---> 17\u001b[0;31m         \u001b[0mgame_state\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mtorch\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfrom_numpy\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mgame_state\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mfloat\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mcuda\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m     18\u001b[0m         \u001b[0;32mreturn\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mlayers\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0mgame_state\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n",
      "\u001b[0;31mKeyboardInterrupt\u001b[0m: "
     ]
    }
   ],
   "source": [
    "model = RankPredictor().cuda().train()\n",
    "opt = optim.Adam(model.parameters(), lr=1e-2)\n",
    "loss = nn.MSELoss()\n",
    "epochs = 10000\n",
    "epoch_losses = []\n",
    "accuracy = []\n",
    "\n",
    "\n",
    "output_tensor = torch.from_numpy(output_train).float().cuda()\n",
    "\n",
    "for epoch in range(epochs):\n",
    "    opt.zero_grad()\n",
    "    predicted = model(input_train)\n",
    "    loss_val = loss(predicted, output_tensor)\n",
    "    loss_val.backward()\n",
    "    opt.step()\n",
    "    \n",
    "    predicted = model(input_test)\n",
    "    predicted = predicted.cpu().detach().numpy().astype(float)\n",
    "#     print(predicted.shape, output_test.shape)\n",
    "#     assert(predicted.shape == output_test.shape)#, \"{}, {}\".format(predicted.shape, output_test.shape))\n",
    "    acc = (abs(predicted.astype(float) - output_test.astype(float)) < 3).sum() / predicted.shape[0]\n",
    "    epoch_losses.append(loss_val.cpu().data.numpy())\n",
    "    accuracy.append(acc)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.plot(range(len(epoch_losses)), epoch_losses);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "plt.plot(range(len(accuracy)), accuracy);"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "d = []\n",
    "for x in range(0, len(input_test), 10):\n",
    "    d.append(model(input_test[x: x+100].astype(float)).mean())\n",
    "    \n",
    "plt.plot(np.array(range(len(d))) / 50 + 10, d)\n",
    "plt.xlabel('Actual rank')\n",
    "plt.ylabel('Predicted rank')\n",
    "plt.xticks(list(range(12, 19)))\n",
    "plt.yticks(list(range(12, 19)));\n",
    "plt.grid();"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "input.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "accuracy[-1]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Ensemble Model"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 618,
   "metadata": {},
   "outputs": [],
   "source": [
    "num_columns = len(input.columns)\n",
    "n_hidden = 32\n",
    "class RankPredictorEnsemble(nn.Module):\n",
    "    def __init__(self):\n",
    "        super().__init__()\n",
    "        self.layers = nn.Sequential(\n",
    "            nn.Linear(num_columns, n_hidden),\n",
    "            nn.ReLU(),\n",
    "            nn.Linear(n_hidden, n_hidden//2),\n",
    "            nn.ReLU(),\n",
    "            nn.Linear(n_hidden//2, 1),\n",
    "            nn.Sigmoid()\n",
    "        )\n",
    "    \n",
    "    def forward(self, game_state):\n",
    "        game_state = torch.from_numpy(game_state).float().cuda()\n",
    "        return self.layers(game_state)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 635,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "67 32\n"
     ]
    }
   ],
   "source": [
    "print(num_columns, n_hidden)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 619,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0, 1, 1, 1, 1, 1]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0, 1, 1, 1, 1, 1]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0, 0, 1, 1, 1, 1]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0, 0, 1, 1, 1, 1]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0, 0, 0, 1, 1, 1]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0, 0, 0, 1, 1, 1]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0, 0, 0, 0, 1, 1]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0, 0, 0, 0, 1, 1]\n"
     ]
    },
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0, 0, 0, 0, 0, 1]\n"
     ]
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAXcAAAEICAYAAACktLTqAAAABHNCSVQICAgIfAhkiAAAAAlwSFlzAAALEgAACxIB0t1+/AAAADl0RVh0U29mdHdhcmUAbWF0cGxvdGxpYiB2ZXJzaW9uIDMuMC4xLCBodHRwOi8vbWF0cGxvdGxpYi5vcmcvDW2N/gAAGSlJREFUeJzt3X2UXPV93/H3dx52Z59XK62EHhEPskFgbNlbDAXXdg0xYAo+rZtC6oObcEJTl9ZNfNIDxy12SHPaPJzEaQ91zElSN44NIU7q6mDVck1wHacFJMKDhcSDEELPaCXtrla7OztP3/4xd5Zhpas7uxpp+Y0+r3P2aO7v/ube3507+sxvfr+5M+buiIhIa0ktdANERKT5FO4iIi1I4S4i0oIU7iIiLUjhLiLSghTuIiItSOEuEjGzr5jZny50O0SaQeEuQTGz3WY2ZWYnzOyQmX3DzLrfBe36uJk9aWZjZrb7FOs/YGZ/Ha3fZ2b/fgGaKecRhbuE6B+4ezfwAWADcP8CtwdgAvhj4Fdj1n8b+DEwAHwU+LyZ3XaO2ibnIYW7BMvdDwGbqYY8AGb2KTN7zsyOm9leM/tK3bq1ZuZm9jkz22NmR8zsS6fatpllzewRM/sLM2troC3PuPs3gV0xVdYC33L3sru/DvwEuKLhgxWZI4W7BMvMVgE3AzvriieAu4B+4FPAvzCzT8+66/XAe4FPAA+Y2eWzttsBfBeYBn7W3Qtmdr2ZjZ5Bc78K3BW9aLwXuBb44RlsT+S0FO4Sou+a2TiwFzgMfLm2wt1/5O4/dfeKu78IPEJ1GKTer7n7lLu/ALwAvL9uXS/wfeB14OfdvRxt9yfu3n8GbX4c+AwwBbwM/JG7bzmD7YmclsJdQvRpd+8BPgZcBiyprTCzD0cTm8NmNgb8Uv36yKG625NA/YTsNcBVwH/yJn2rnpkNUH3BeBDIAauBT5rZ55uxfZFTUbhLsNz9/wDfAH6nrvjbwEZgtbv3AX8A2Bw2+wPgPwJPmNmyJjX1YqDs7n/i7iV33wc8CtzSpO2LnEThLqH7KnCjmdWGVnqAY+6eN7OrgZ+b6wbd/beovkg8YWaze/2nZGYpM8sB2eqi5eomYl+Nyn4uqncB8E+AF+faNpFGKdwlaO4+DPwJ8EBU9HngwWhM/gHgsXlu99epTqr+0MwGzOwjZnbiNHf5e1TH0zcBa6LbP4i2dRz4h8AvAyPA88A24D/Mp20ijTD9WIeISOtRz11EpAUp3EVEWpDCXUSkBSncRURaUGahdrxkyRJfu3btQu1eRCRIzz777BF3H0yqt2DhvnbtWrZu3bpQuxcRCZKZvdlIPQ3LiIi0IIW7iEgLUriLiLQghbuISAtSuIuItCCFu4hIC1K4i4i0oAX7nLtIK8kXy+Sy6ZPKj+eLdGbTpFOGWfxvhpQrjgGpVLWOu8/UL5YrpMxwdzLp1Mw6d+d4vkRvLsPBsTwOLOtpp+Kw++gElwxWf2Dq+FSRYqXC4q520iljPF8knTLa0inK7pTKzuhUkRV9OcoVZ/fRSVb2d5BKMbNueW+ObQfGSJlx8WAXh8by/PVrRxjPF7lqVT/XXrKYqWKZN4YneH7vKJm0cetVKyiUKrx5dIJUyrh0aTfHThQYmSxQrjjj+RLt2RS9uSzDJ6ZZ1NlGoVRhUWeWfLHCjkPHyaSMXcMTXNCXY8fB41y1qo9sOkU6ZRTLzkVLOulsyzA6WQTg5UPHyaRT9OYyrF3cxXi+xP7RSVb0d5AyI18sU/Hq433lyl7G8yXG8yVKlQpLe9ppS6d59a1xPHrcMynjlUPjLO5u5/LlPfR3tlGuOIM97bxxZIIrVvSSTacoV5yUMXPOdh4ep1ByVvTn6O9sI18sU644f7tnhInpEh+/bCntmZOfL82kcJczVntiTxXLtGfSlCoV2jPpd4QQQL5YIZet/keYLJbJplKMTRV58+gEa5d0MTpZ5NhEgUK5grszVSjTk8vy6lvjfPyypYxOFhibKtLdXg2zvSOTXLyki+ETBfaNTLJmoJOObJrxfImu9gwHRqeYLJTp78wyMlHgxHSJikNvR4bJ6TJjU0X6OrLksinePDrJD7a/xcr+Dq6/dAl/tnUvt161HAdK5Qo9uSz5Ypm9I1McPp7nsgt6ODpR4MV9Y+94LN6/up+lPe28+tY4qxZ18MwbxyiWna62NL0dWU5Ml8hlq4/N+hV97Ds2yeqBTv52zwjj+RIr+nJMFcsUShX6OrIcGMuf9Hh3t2c4MV06F6f2jHzpf2xr+ja/9XTTN3lGMilj1aIOdh+dBGCgq42RyQK1b1LvakuTzaRmXnxq7rv5Mn7po5ec1bYt2Pe5Dw0Nua5QPTsqFa+G5rFJyhVnZKLA83tHOTyeZ3F3O1/70esArOjLcWAszxUrejk0lmfN4k7WLu5ioKuNjqgXeng8z4r+DnYePsFfvXyYSwa7+en+MTIpo1QJ77cAMimj7M7pnvb1x3ZBb45SxZkslBjoasMdDoxNsbirnelimfG6kL18eS/TxTLTpQoAgz3tPL93lPcs62Zo7QCjkwW62jJMFEp8f9shKl7dVyZt5IvVXuKlS7tpz6Roy6RY0t3OeL7E6FSBYycKHBjLc+HiTq5c0cfoVIG/2Xl0Zt8ff+8gz+0dZXSy2gPfcOEivvfiQQBuuHwZV67s5as/fA2AW69azvF8iR+/Ojxz/89es4Y/fWoPAHddeyFP7DjM/tEp/vlHLwaHbz71Jh9Zt4Tr1w3yG9/bTr5Y4a5rL6SvI8uizjZ2HDzOZKFMR1uavo4sh8enWdSZZVlvjonp6ott9cU6Q8WhvzPLxucP0JZJcdkFPVwy2E1bJkWp4rg7bZlU9MKbZs/RSfaNTNLZnqFYqrCsL8eJfImKO8enimx9c4Q1A518ZN0gvR0ZHn/hIOm08aE1i9g7Mskjz+zhhsuXcUFvjmK5wvCJAos6s/TksuwaPsHRiQKfvGIZk4Uy7sy0d3SywNLeHCMTBXpyGcoOk9Mlnt0zgjts3X2MT29YyWuHT/DMG8cAWNzVRi6bZmV/Bx9Zt4QX9o3x9K6j9OQyrF/Ry3WXLmGgq42/e8kSBnva5/UcNrNn3X0osZ7CPXzb9o/xf18/woHRPPtGJtn65shJPQUzyKZSFMqV026rLZMimzKmorevZ6I3l2F8usSH1iziksFunnj5MB1tKfYem2LDmn6K5QpThTJd7Rku6M2x49BxenPVXvbi7namimWW9bZzZLzAPx5aRTpl0X9AJ5tOMdjTzoWLuzhyYprFXW20Z9IUyhVWD3Qwni/Rlq6+5S+UK7RnU6TMqmGasplhkmK5Er3zMI5OTLOsJ0cqZUxFQXW2TBWqvfPejuqb59MN2ZxNx/NFenPZBdm3zE+j4a5hmcC9dGCMW//LT4BqMHe1pfnQmkVsWNPPmsVd9OQytKdTvG9VH11tGSaLZVIGxbLTnkmxf3SKtYu7KFUqtKVTMyHj7pyYLs3cJ5dJcXAsz7LeHMcmCmTTxkBX9SdC6++zUCE129Kexupl0ylqQ+XL+zpmys9msNe2f7b30QgFe+sKMtx3H5lgy+5jlAMcFmi25/eOArDx3ut438q+xHDtbn/nKa9NuqVT7wwaM6Mn+o9fu8/qgU4ALujLnXLb75ZgF5FAw/0X/vsWdg1PLHQz3jWWdLdzxYrkYBeR80dw4T5VKPPGkQnu+Dur+cIN6xa6Oe8KfR1Z0ikFu4i8LbhwPzg2hTt8+OKBd4yRiojI24K7QrU2zJ5OBdd0EZFzJsCE1CSqiEiSAMO9SiPMIiLxgg13ERGJF1y4L9AFtSIiQQku3Gv0kW4RkXjBhbs67iIiyYIL9xrTlKqISKxgw11EROIFF+6aUBURSRZcuNdoQlVEJF5w4e6aUhURSRRcuNeo4y4iEi/YcBcRkXjBhbsmVEVEkgUX7jWaUBURiRdcuKvnLiKSLLhwf5u67iIicRoKdzO7ycxeMbOdZnbfKdavMbMnzew5M3vRzG5pflOr9FFIEZFkieFuZmngIeBmYD1wp5mtn1Xt3wGPufsG4A7gvza7oSIi0rhGeu5XAzvdfZe7F4BHgdtn1XGgN7rdBxxoXhNPTROqIiLxGgn3lcDeuuV9UVm9rwCfNbN9wCbgX51qQ2Z2j5ltNbOtw8PD82iuJlRFRBrRrAnVO4FvuPsq4Bbgm2Z20rbd/WF3H3L3ocHBwTPaoTruIiLxGgn3/cDquuVVUVm9u4HHANz9/wE5YEkzGigiInPXSLhvAdaZ2UVm1kZ1wnTjrDp7gE8AmNnlVMN9fuMuDTINuouIxEoMd3cvAfcCm4EdVD8V85KZPWhmt0XVvgj8opm9ADwC/DN3jY6LiCyUTCOV3H0T1YnS+rIH6m5vB65rbtPi2nIu9iIiErZgr1DVoIyISLzgwl1XqIqIJAsu3Gs0nyoiEi/YcBcRkXjBhbsmVEVEkgUX7jUalhERiRdcuKvjLiKSLLhwrzF9GFJEJFaw4S4iIvGCC3d9q4GISLLgwn2GRmVERGIFF+7qt4uIJAsu3GvUcRcRiRdsuIuISLzgwl3zqSIiyYIL9xr9EpOISLwAw11ddxGRJAGGe5X67SIi8YILd425i4gkCy7cazTkLiISL9hwFxGReMGFu0ZlRESSBRfuNfrKXxGReMGFuyZURUSSBRfuNZpQFRGJF2y4i4hIvODCXT/WISKSLLhwr9GojIhIvODCXf12EZFkwYX7DHXdRURihRvuIiISK7hw13yqiEiy4MK9RleoiojECy7cXVOqIiKJggv3Gl2hKiISL9hwFxGReOGFu0ZlREQShRfuEY3KiIjECy7c1XEXEUnWULib2U1m9oqZ7TSz+2Lq/KyZbTezl8zs281t5in3d7Z3ISISrExSBTNLAw8BNwL7gC1mttHdt9fVWQfcD1zn7iNmtvRsNVhERJI10nO/Gtjp7rvcvQA8Ctw+q84vAg+5+wiAux9ubjPfpitURUSSNRLuK4G9dcv7orJ67wHeY2Z/Y2ZPmdlNzWpgHI3KiIjESxyWmcN21gEfA1YBPzaz97n7aH0lM7sHuAdgzZo189qRrlAVEUnWSM99P7C6bnlVVFZvH7DR3Yvu/gbwKtWwfwd3f9jdh9x9aHBwcL5tBvRRSBGR02kk3LcA68zsIjNrA+4ANs6q812qvXbMbAnVYZpdTWznDI25i4gkSwx3dy8B9wKbgR3AY+7+kpk9aGa3RdU2A0fNbDvwJPCr7n70bDUaNOYuInI6DY25u/smYNOssgfqbjvwK9GfiIgsMF2hKiLSgoIL97dpXEZEJE5w4e6aURURSRRcuNdoQlVEJF6w4S4iIvGCC3cNyoiIJAsu3Gs0KiMiEi+8cFfXXUQkUXjhHtGPdYiIxAs23EVEJF5w4a6v/BURSRZcuNdoUEZEJF5w4a4LVEVEkgUX7jWaTxURiRdsuIuISLzgwl3DMiIiyYIL9xrTlKqISKzgwl0ddxGRZMGFe40mVEVE4gUX7vqxDhGRZMGFu4iIJFO4i4i0oODCXYMyIiLJggv3Gk2oiojECy7cNZ8qIpIsuHCv0UVMIiLxgg13ERGJF2C4a1xGRCRJgOFepQlVEZF4wYW7JlRFRJIFF+416rmLiMQLNtxFRCRecOGuURkRkWTBhXuNPucuIhIvuHDXhKqISLLgwr1GE6oiIvGCDXcREYkXXLi7plRFRBIFF+41GpUREYkXXLhrQlVEJFlD4W5mN5nZK2a208zuO029f2RmbmZDzWti3L7O9h5ERMKVGO5mlgYeAm4G1gN3mtn6U9TrAb4APN3sRoqIyNw00nO/Gtjp7rvcvQA8Ctx+inq/DvwmkG9i+06iURkRkWSNhPtKYG/d8r6obIaZfRBY7e7fO92GzOweM9tqZluHh4fn3NhZWzvD+4uItK4znlA1sxTwu8AXk+q6+8PuPuTuQ4ODg/Pan2tGVUQkUSPhvh9YXbe8Kiqr6QGuBH5kZruBa4CNZ3tSVROqIiLxGgn3LcA6M7vIzNqAO4CNtZXuPubuS9x9rbuvBZ4CbnP3rWelxSIikigx3N29BNwLbAZ2AI+5+0tm9qCZ3Xa2GygiInOXaaSSu28CNs0qeyCm7sfOvFnJNCojIhJPV6iKiLSg4MK9xjSjKiISK7hw17dCiogkCy7ca9RvFxGJF2y4i4hIvODCXROqIiLJggv3Gs2niojECy7c1XMXEUkWXLjXmKZURURiBRvuIiISL7hw16iMiEiy4MK9RhOqIiLxggt3/ViHiEiy4MJdRESSKdxFRFpQcOGuQRkRkWTBhXuNJlRFROKFF+7quouIJAov3CP6sQ4RkXjBhruIiMQLLtz1S0wiIsmCC/caDcqIiMQLLtx1gaqISLLgwr1G86kiIvGCC3d13EVEkgUX7jX6sQ4RkXjBhruIiMQLLtw1oSoikiy4cK/RhKqISLzgwl0XMYmIJAsu3GvUcRcRiRdsuIuISLzgwl0TqiIiyYIL9xkalxERiRVcuKvjLiKSLLhwr9EVqiIi8YINdxERiRdeuGtGVUQkUXjhHtEVqiIi8RoKdzO7ycxeMbOdZnbfKdb/ipltN7MXzewJM7uw+U2tUr9dRCRZYribWRp4CLgZWA/caWbrZ1V7Dhhy96uA7wC/1eyGntSus70DEZGANdJzvxrY6e673L0APArcXl/B3Z9098lo8SlgVXObKSIic9FIuK8E9tYt74vK4twN/K9TrTCze8xsq5ltHR4ebryVdTSfKiKSrKkTqmb2WWAI+O1TrXf3h919yN2HBgcHz3RfZ3R/EZFWlmmgzn5gdd3yqqjsHczsBuBLwEfdfbo5zTuZq+suIpKokZ77FmCdmV1kZm3AHcDG+gpmtgH4OnCbux9ufjNPpn67iEi8xHB39xJwL7AZ2AE85u4vmdmDZnZbVO23gW7gz83seTPbGLM5ERE5BxoZlsHdNwGbZpU9UHf7hia3K74t52pHIiIB0xWqIiItKLhw13yqiEiy4MK9Rl/5KyISL7hwV8ddRCRZcOE+Qx13EZFY4Ya7iIjECi7cdYWqiEiy4MK9Rh+FFBGJF2y4i4hIvGDDXR13EZF4wYa7iIjECy7cNZ8qIpIsuHCv0Y91iIjECy7cXdeoiogkCi7ca9RvFxGJF2y4i4hIvODCXROqIiLJggv3Gs2niojECy7cLx7s5lPvW05K6S4iEquh31B9N7lx/TJuXL9soZshIvKuFlzPXUREkincRURakMJdRKQFKdxFRFqQwl1EpAUp3EVEWpDCXUSkBSncRURakPkCfVmLmQ0Db87z7kuAI01sTgh0zOcHHfP54UyO+UJ3H0yqtGDhfibMbKu7Dy10O84lHfP5Qcd8fjgXx6xhGRGRFqRwFxFpQaGG+8ML3YAFoGM+P+iYzw9n/ZiDHHMXEZHTC7XnLiIip6FwFxFpQcGFu5ndZGavmNlOM7tvodvTLGa22syeNLPtZvaSmX0hKh8ws/9tZq9F/y6Kys3M/nP0OLxoZh9c2COYHzNLm9lzZvZ4tHyRmT0dHdefmVlbVN4eLe+M1q9dyHbPl5n1m9l3zOxlM9thZteeB+f4l6Pn9DYze8TMcq14ns3sj83ssJltqyub87k1s89F9V8zs8/Ntz1BhbuZpYGHgJuB9cCdZrZ+YVvVNCXgi+6+HrgG+JfRsd0HPOHu64AnomWoPgbror97gK+d+yY3xReAHXXLvwn8nrtfCowAd0fldwMjUfnvRfVC9PvA9939MuD9VI+9Zc+xma0E/jUw5O5XAmngDlrzPH8DuGlW2ZzOrZkNAF8GPgxcDXy59oIwZ+4ezB9wLbC5bvl+4P6FbtdZOtb/CdwIvAIsj8qWA69Et78O3FlXf6ZeKH/AqugJ//eBxwGjetVeZvb5BjYD10a3M1E9W+hjmOPx9gFvzG53i5/jlcBeYCA6b48Dn2zV8wysBbbN99wCdwJfryt/R725/AXVc+ftJ0rNvqispURvRTcATwPL3P1gtOoQUPsB2VZ4LL4K/FugEi0vBkbdvRQt1x/TzPFG68ei+iG5CBgG/ls0FPWHZtZFC59jd98P/A6wBzhI9bw9S2uf53pzPbdNO+ehhXvLM7Nu4C+Af+Pux+vXefWlvCU+u2pmtwKH3f3ZhW7LOZQBPgh8zd03ABO8/TYdaK1zDBANKdxO9YVtBdDFyUMX54VzfW5DC/f9wOq65VVRWUswsyzVYP+Wu/9lVPyWmS2P1i8HDkfloT8W1wG3mdlu4FGqQzO/D/SbWSaqU39MM8cbre8Djp7LBjfBPmCfuz8dLX+Hati36jkGuAF4w92H3b0I/CXVc9/K57neXM9t0855aOG+BVgXzbS3UZ2Y2bjAbWoKMzPgj4Ad7v67das2ArUZ889RHYuvld8VzbpfA4zVvf1713P3+919lbuvpXoe/8rd/ynwJPCZqNrs4609Dp+J6gfVw3X3Q8BeM3tvVPQJYDsteo4je4BrzKwzeo7Xjrllz/Mscz23m4GfMbNF0buen4nK5m6hJyDmMWFxC/Aq8DrwpYVuTxOP63qqb9leBJ6P/m6hOt74BPAa8ENgIKpvVD859DrwU6qfRljw45jnsX8MeDy6fTHwDLAT+HOgPSrPRcs7o/UXL3S753msHwC2Ruf5u8CiVj/HwK8BLwPbgG8C7a14noFHqM4rFKm+S7t7PucW+IXo+HcCPz/f9ujrB0REWlBowzIiItIAhbuISAtSuIuItCCFu4hIC1K4i4i0IIW7iEgLUriLiLSg/w/yqLMitVSN9gAAAABJRU5ErkJggg==\n",
      "text/plain": [
       "<Figure size 432x288 with 1 Axes>"
      ]
     },
     "metadata": {
      "needs_background": "light"
     },
     "output_type": "display_data"
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0, 0, 0, 0, 0, 0]\n"
     ]
    },
    {
     "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": [
    "ranks = list(range(10, 20))\n",
    "models = {}\n",
    "for rank in ranks:\n",
    "    models[rank] = RankPredictorEnsemble().cuda()\n",
    "for rank, model in models.items():\n",
    "#     if rank != 15:\n",
    "#         continue\n",
    "    opt = optim.Adam(model.parameters(), lr=1e-2)\n",
    "    loss = nn.BCELoss()\n",
    "    epochs = 1000\n",
    "    epoch_losses = []\n",
    "    accuracy = []\n",
    "\n",
    "    out = (output_train > rank).astype(int)\n",
    "    print([out[n][0] for n in range(0, len(output_train), len(output_train)//5)])\n",
    "    output_tensor = torch.from_numpy(out).cuda().float()\n",
    "    for epoch in range(epochs):\n",
    "        opt.zero_grad()\n",
    "        predicted = model(input_train)\n",
    "        loss_val = loss(predicted, output_tensor)\n",
    "        loss_val.backward()\n",
    "        opt.step()\n",
    "\n",
    "        predicted = model(input_test)\n",
    "        predicted = predicted.cpu().detach().numpy().astype(float)\n",
    "    #     print(predicted.shape, output_test.shape)\n",
    "    #     assert(predicted.shape == output_test.shape)#, \"{}, {}\".format(predicted.shape, output_test.shape))\n",
    "        acc = ((predicted > 0.5).astype(int) == (output_test > rank).astype(int)).sum() / predicted.shape[0]\n",
    "        epoch_losses.append(loss_val.cpu().data.numpy())\n",
    "        accuracy.append(acc)\n",
    "    plt.plot(range(len(accuracy)), accuracy);\n",
    "    plt.title('Rank: {}'.format(rank));\n",
    "    plt.show();"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 620,
   "metadata": {},
   "outputs": [],
   "source": [
    "df_test = None\n",
    "for rank, model in models.items():\n",
    "    predicted = pd.Series(model(input_test).detach().cpu().numpy().reshape(len(input_test)), name=str(rank))\n",
    "    if df_test is None:\n",
    "        df_test = predicted.to_frame()\n",
    "    else:\n",
    "        df_test = df_test.merge(predicted.to_frame(), left_index=True, right_index=True)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 621,
   "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>10</th>\n",
       "      <th>11</th>\n",
       "      <th>12</th>\n",
       "      <th>13</th>\n",
       "      <th>14</th>\n",
       "      <th>15</th>\n",
       "      <th>16</th>\n",
       "      <th>17</th>\n",
       "      <th>18</th>\n",
       "      <th>19</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.993729</td>\n",
       "      <td>0.974568</td>\n",
       "      <td>0.983082</td>\n",
       "      <td>0.898597</td>\n",
       "      <td>0.867083</td>\n",
       "      <td>0.493820</td>\n",
       "      <td>0.400019</td>\n",
       "      <td>0.142242</td>\n",
       "      <td>0.074801</td>\n",
       "      <td>1.148884e-10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.855384</td>\n",
       "      <td>0.714060</td>\n",
       "      <td>0.437516</td>\n",
       "      <td>0.239995</td>\n",
       "      <td>0.193985</td>\n",
       "      <td>0.029841</td>\n",
       "      <td>0.021862</td>\n",
       "      <td>0.005633</td>\n",
       "      <td>0.003909</td>\n",
       "      <td>2.761742e-11</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.773590</td>\n",
       "      <td>0.646934</td>\n",
       "      <td>0.352010</td>\n",
       "      <td>0.123321</td>\n",
       "      <td>0.119133</td>\n",
       "      <td>0.142647</td>\n",
       "      <td>0.088494</td>\n",
       "      <td>0.026729</td>\n",
       "      <td>0.012820</td>\n",
       "      <td>3.032887e-10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.511579</td>\n",
       "      <td>0.153341</td>\n",
       "      <td>0.087233</td>\n",
       "      <td>0.043878</td>\n",
       "      <td>0.001956</td>\n",
       "      <td>0.005382</td>\n",
       "      <td>0.002470</td>\n",
       "      <td>0.000639</td>\n",
       "      <td>0.000074</td>\n",
       "      <td>1.532744e-10</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.857064</td>\n",
       "      <td>0.805974</td>\n",
       "      <td>0.438387</td>\n",
       "      <td>0.268981</td>\n",
       "      <td>0.065997</td>\n",
       "      <td>0.129498</td>\n",
       "      <td>0.036683</td>\n",
       "      <td>0.025293</td>\n",
       "      <td>0.004045</td>\n",
       "      <td>7.668416e-10</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "         10        11        12        13        14        15        16  \\\n",
       "0  0.993729  0.974568  0.983082  0.898597  0.867083  0.493820  0.400019   \n",
       "1  0.855384  0.714060  0.437516  0.239995  0.193985  0.029841  0.021862   \n",
       "2  0.773590  0.646934  0.352010  0.123321  0.119133  0.142647  0.088494   \n",
       "3  0.511579  0.153341  0.087233  0.043878  0.001956  0.005382  0.002470   \n",
       "4  0.857064  0.805974  0.438387  0.268981  0.065997  0.129498  0.036683   \n",
       "\n",
       "         17        18            19  \n",
       "0  0.142242  0.074801  1.148884e-10  \n",
       "1  0.005633  0.003909  2.761742e-11  \n",
       "2  0.026729  0.012820  3.032887e-10  \n",
       "3  0.000639  0.000074  1.532744e-10  \n",
       "4  0.025293  0.004045  7.668416e-10  "
      ]
     },
     "execution_count": 621,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df_test.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 622,
   "metadata": {},
   "outputs": [],
   "source": [
    "results = df_test.apply(lambda x: x > 0.5).apply(lambda x: int(x.idxmin()), axis=1).values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 623,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'Predicted Rank')"
      ]
     },
     "execution_count": 623,
     "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.scatter(output_test.reshape(output_test.shape[0]), results, alpha=0.005)\n",
    "plt.xlabel('Actual Rank')\n",
    "plt.ylabel('Predicted Rank')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 624,
   "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>Accuracy (%)</th>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>Rank Difference</th>\n",
       "      <th></th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>23.087594</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>60.860087</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>82.065002</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>93.182719</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>97.502973</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>99.385652</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>99.821641</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>99.960365</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>100.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>100.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>100.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>100.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>100.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>100.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>100.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "                 Accuracy (%)\n",
       "Rank Difference              \n",
       "0                   23.087594\n",
       "1                   60.860087\n",
       "2                   82.065002\n",
       "3                   93.182719\n",
       "4                   97.502973\n",
       "5                   99.385652\n",
       "6                   99.821641\n",
       "7                   99.960365\n",
       "8                  100.000000\n",
       "9                  100.000000\n",
       "10                 100.000000\n",
       "11                 100.000000\n",
       "12                 100.000000\n",
       "13                 100.000000\n",
       "14                 100.000000"
      ]
     },
     "execution_count": 624,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "d = []\n",
    "for i in range(15):\n",
    "    d.append((abs(results - output_test.reshape(output_test.shape[0])) <= i).sum() * 100 / results.shape[0])\n",
    "d = pd.DataFrame(d, columns = ['Accuracy (%)'])\n",
    "d.index.name = 'Rank Difference'\n",
    "d"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 652,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['goals',\n",
       " 'assists',\n",
       " 'saves',\n",
       " 'shots',\n",
       " 'field_of_view',\n",
       " 'transition_speed',\n",
       " 'pitch',\n",
       " 'swivel_speed',\n",
       " 'stiffness',\n",
       " 'height',\n",
       " 'distance',\n",
       " 'total_hits',\n",
       " 'total_dribbles',\n",
       " 'total_dribble_conts',\n",
       " 'total_goals',\n",
       " 'total_shots',\n",
       " 'total_passes',\n",
       " 'total_aerials',\n",
       " 'turnovers',\n",
       " 'turnovers_on_my_half',\n",
       " 'turnovers_on_their_half',\n",
       " 'won_turnovers',\n",
       " 'possession_time',\n",
       " 'average_speed',\n",
       " 'average_hit_distance',\n",
       " 'average_distance_from_center',\n",
       " 'boost_usage',\n",
       " 'num_small_boosts',\n",
       " 'num_large_boosts',\n",
       " 'num_stolen_boosts',\n",
       " 'wasted_collection',\n",
       " 'wasted_usage',\n",
       " 'time_full_boost',\n",
       " 'time_low_boost',\n",
       " 'time_no_boost',\n",
       " 'average_boost_level',\n",
       " 'wasted_big',\n",
       " 'wasted_small',\n",
       " 'time_on_ground',\n",
       " 'time_low_in_air',\n",
       " 'time_high_in_air',\n",
       " 'time_in_defending_half',\n",
       " 'time_in_attacking_half',\n",
       " 'time_in_defending_third',\n",
       " 'time_in_neutral_third',\n",
       " 'time_in_attacking_third',\n",
       " 'time_behind_ball',\n",
       " 'time_in_front_ball',\n",
       " 'time_closest_to_ball',\n",
       " 'time_furthest_from_ball',\n",
       " 'time_close_to_ball',\n",
       " 'time_near_wall',\n",
       " 'time_in_corner',\n",
       " 'ball_hit_forward',\n",
       " 'ball_hit_backward',\n",
       " 'is_keyboard',\n",
       " 'time_at_boost_speed',\n",
       " 'time_at_slow_speed',\n",
       " 'time_at_super_sonic',\n",
       " 'time_closest_to_team_center',\n",
       " 'time_furthest_from_team_center',\n",
       " 'first_frame_in_game',\n",
       " 'time_in_front_of_center_of_mass',\n",
       " 'time_behind_center_of_mass',\n",
       " 'time_most_forward_player',\n",
       " 'time_most_back_player',\n",
       " 'time_between_players']"
      ]
     },
     "execution_count": 652,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "list(input.columns)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 650,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "# os.mkdir('models')\n",
    "for rank, model in models.items():\n",
    "    with open(os.path.join('models', '{}.mdl'.format(rank)), 'wb') as f:\n",
    "        torch.save(model.state_dict(), f)\n",
    "maxs.to_csv(os.path.join('models', 'maxs.csv'))\n",
    "mins.to_csv(os.path.join('models', 'mins.csv'))\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 645,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0\n",
       "goals             0.490253\n",
       "assists           0.392319\n",
       "saves             0.307938\n",
       "shots             0.980507\n",
       "field_of_view    20.591298\n",
       "Name: 1, dtype: float64"
      ]
     },
     "execution_count": 645,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "pd.read_csv('models/maxs.csv', index_col=0, squeeze=True, header=None).head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 648,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "82"
      ]
     },
     "execution_count": 648,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "len(smurf_data.values[0])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 625,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "https://calculated.gg/api/v1/playergames?key=1&player_ids=76561197992682099&playlist=13&days=30\n"
     ]
    },
    {
     "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>index</th>\n",
       "      <th>id</th>\n",
       "      <th>name</th>\n",
       "      <th>player</th>\n",
       "      <th>game</th>\n",
       "      <th>rank</th>\n",
       "      <th>division</th>\n",
       "      <th>mmr</th>\n",
       "      <th>is_orange</th>\n",
       "      <th>win</th>\n",
       "      <th>...</th>\n",
       "      <th>time_closest_to_team_center</th>\n",
       "      <th>time_furthest_from_team_center</th>\n",
       "      <th>is_bot</th>\n",
       "      <th>first_frame_in_game</th>\n",
       "      <th>time_in_game</th>\n",
       "      <th>time_in_front_of_center_of_mass</th>\n",
       "      <th>time_behind_center_of_mass</th>\n",
       "      <th>time_most_forward_player</th>\n",
       "      <th>time_most_back_player</th>\n",
       "      <th>time_between_players</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0</td>\n",
       "      <td>1810312</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>F42DF61811E8D0AFB57A4CBBE67B0A36</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1712</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>316.817078</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>2280047</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>99812C0A11E8D7296F926E8B15A8A97B</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1645</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>90.470673</td>\n",
       "      <td>23.071768</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>320.870453</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2</td>\n",
       "      <td>2310713</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>68A4B98011E8D87D5CEB60A690E3C53A</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1862</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>309.791565</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>3</td>\n",
       "      <td>2433067</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>C794499E11E8DD3FD67A2DB278296A08</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1932</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>239.715698</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>4</td>\n",
       "      <td>2279879</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>A0EB8E3211E8D72888C06E8BA264708C</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1645</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>71.419067</td>\n",
       "      <td>15.992106</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>304.897400</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>5</th>\n",
       "      <td>5</td>\n",
       "      <td>2448565</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>16DDCCFA11E8CF444EA4C3A195DACF62</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1561</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>73.591843</td>\n",
       "      <td>12.885868</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>317.122864</td>\n",
       "      <td>129.079178</td>\n",
       "      <td>188.043671</td>\n",
       "      <td>93.191521</td>\n",
       "      <td>136.420151</td>\n",
       "      <td>87.511192</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>6</th>\n",
       "      <td>6</td>\n",
       "      <td>1982626</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>0797E9E011E8D2433E145E963BE58303</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1745</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>267.241119</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>7</th>\n",
       "      <td>7</td>\n",
       "      <td>2351858</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>C709B9D811E8D89888C04BABA264708C</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1684</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>59.694107</td>\n",
       "      <td>14.634933</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>322.558838</td>\n",
       "      <td>142.799973</td>\n",
       "      <td>179.758850</td>\n",
       "      <td>103.126450</td>\n",
       "      <td>122.617470</td>\n",
       "      <td>96.814919</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>8</th>\n",
       "      <td>8</td>\n",
       "      <td>2311313</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>93F3FF7A11E8D8810833C280918F6189</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1862</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>279.595490</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>9</th>\n",
       "      <td>9</td>\n",
       "      <td>2452756</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>4EF593E811E8D1BF003EF2957E06B85B</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1706</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>64.702675</td>\n",
       "      <td>14.938508</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>307.896698</td>\n",
       "      <td>156.848862</td>\n",
       "      <td>151.047836</td>\n",
       "      <td>112.929527</td>\n",
       "      <td>91.868462</td>\n",
       "      <td>103.098709</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>10</th>\n",
       "      <td>10</td>\n",
       "      <td>2414799</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>F4A7E8C011E8DC7FBB623D9E327FF865</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1703</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>136.537918</td>\n",
       "      <td>39.434074</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>559.563354</td>\n",
       "      <td>206.932556</td>\n",
       "      <td>352.593109</td>\n",
       "      <td>122.960953</td>\n",
       "      <td>259.256866</td>\n",
       "      <td>177.345535</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>11</th>\n",
       "      <td>11</td>\n",
       "      <td>2414903</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>6A4FEB6211E8DC815953989D60D4B411</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1703</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>78.854057</td>\n",
       "      <td>19.491844</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>309.838928</td>\n",
       "      <td>148.863434</td>\n",
       "      <td>160.975494</td>\n",
       "      <td>99.109039</td>\n",
       "      <td>116.117851</td>\n",
       "      <td>94.612038</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>12</th>\n",
       "      <td>12</td>\n",
       "      <td>2433581</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>8A02699E11E8DD4447B6FBB5D3AEA3D3</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1915</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>191.794312</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>13</th>\n",
       "      <td>13</td>\n",
       "      <td>1498810</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>AE01BB0211E8C9AC87EF4A91777CAF14</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1558</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>76.840515</td>\n",
       "      <td>7.734396</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>320.613831</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>14</th>\n",
       "      <td>14</td>\n",
       "      <td>2414556</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>F56EB3FC11E8DC7EA5D4E2A31FFD83A0</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1703</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>80.277771</td>\n",
       "      <td>20.290960</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>305.823883</td>\n",
       "      <td>110.405708</td>\n",
       "      <td>195.418182</td>\n",
       "      <td>63.209217</td>\n",
       "      <td>137.904694</td>\n",
       "      <td>104.709976</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>15</th>\n",
       "      <td>15</td>\n",
       "      <td>2312425</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>9FC51B8E11E8D888396DFB8A68244BC6</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1868</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>339.367737</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>16</th>\n",
       "      <td>16</td>\n",
       "      <td>1498379</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>AF3DA2A011E8C9ADD67A4A9178296A08</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1558</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>81.425682</td>\n",
       "      <td>14.498387</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>311.646606</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>17</th>\n",
       "      <td>17</td>\n",
       "      <td>1498696</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>C042DD8011E8C9AEF5C44A91BCD7229C</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1558</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>72.720779</td>\n",
       "      <td>7.823503</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>311.937500</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>18</th>\n",
       "      <td>18</td>\n",
       "      <td>2356005</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>8634AE0E11E8D9AB0FD7BD9D1B760370</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1684</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>52.461136</td>\n",
       "      <td>21.522015</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>308.718506</td>\n",
       "      <td>141.135162</td>\n",
       "      <td>167.583359</td>\n",
       "      <td>99.395424</td>\n",
       "      <td>118.196808</td>\n",
       "      <td>91.126274</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>19</th>\n",
       "      <td>19</td>\n",
       "      <td>2280307</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>8FE6C37E11E8D72BF6B26E8B705CB0BC</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1645</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>79.470146</td>\n",
       "      <td>44.426254</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>358.820129</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>20</th>\n",
       "      <td>20</td>\n",
       "      <td>1926656</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>737245B211E8D1C01F441D93CE788AEF</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1570</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>76.285492</td>\n",
       "      <td>25.516151</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>316.683105</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>21</th>\n",
       "      <td>21</td>\n",
       "      <td>2356023</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>834B417011E8D9ACCB19AD8ADEE4DFF5</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1684</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>95.315987</td>\n",
       "      <td>22.648022</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>338.979065</td>\n",
       "      <td>155.768936</td>\n",
       "      <td>183.100800</td>\n",
       "      <td>97.884315</td>\n",
       "      <td>127.128174</td>\n",
       "      <td>113.966576</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>22</th>\n",
       "      <td>22</td>\n",
       "      <td>1984368</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>1250A80011E8D247E993E3A0E1517F09</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1734</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>323.215607</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>23</th>\n",
       "      <td>23</td>\n",
       "      <td>2280591</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>AC25F35A11E8D72D7B426E8B6A3A4A60</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1645</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>72.441048</td>\n",
       "      <td>20.536903</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>319.386444</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>24</th>\n",
       "      <td>24</td>\n",
       "      <td>1498995</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>CFADFCC211E8C9AF66EB4A919DB5566D</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1558</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>74.547325</td>\n",
       "      <td>25.207628</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>314.491699</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>25</th>\n",
       "      <td>25</td>\n",
       "      <td>2020751</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>0FC4A22C11E8D29B1D9496B17D913D63</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1625</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>84.157669</td>\n",
       "      <td>25.863363</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>315.624603</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>26</th>\n",
       "      <td>26</td>\n",
       "      <td>1985622</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>C0DF35BA11E8D249E6399B9BDD4705C1</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1734</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>362.147217</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>27</th>\n",
       "      <td>27</td>\n",
       "      <td>2311897</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>C2475BE411E8D8847EA2F3941DE3A62F</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1868</td>\n",
       "      <td>True</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>416.602844</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>28</th>\n",
       "      <td>28</td>\n",
       "      <td>2520592</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>710CEACC11E8DFF6CD02CDA535F4529D</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1737</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>64.296173</td>\n",
       "      <td>13.021171</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>315.299683</td>\n",
       "      <td>155.757812</td>\n",
       "      <td>159.541870</td>\n",
       "      <td>108.971848</td>\n",
       "      <td>109.194321</td>\n",
       "      <td>97.133514</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>29</th>\n",
       "      <td>29</td>\n",
       "      <td>2326146</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>9D45BB9E11E8D8F07ABAEE9FE3DA7507</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1908</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>325.902771</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>30</th>\n",
       "      <td>30</td>\n",
       "      <td>1982187</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>E332DB4211E8D241F5C4B092BCD7229C</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1745</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>345.010986</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>31</th>\n",
       "      <td>31</td>\n",
       "      <td>2415056</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>4E0F1DC811E8DC828ACD30A07597FDD4</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1703</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>83.874092</td>\n",
       "      <td>20.201830</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>357.592102</td>\n",
       "      <td>187.077820</td>\n",
       "      <td>170.514282</td>\n",
       "      <td>121.785156</td>\n",
       "      <td>124.780144</td>\n",
       "      <td>111.026802</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>32</th>\n",
       "      <td>32</td>\n",
       "      <td>1498894</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>834EE2F011E8C9ABD5514A91DFFAAFCB</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1558</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>71.580940</td>\n",
       "      <td>17.121521</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>315.177673</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>33</th>\n",
       "      <td>33</td>\n",
       "      <td>2074626</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>C0F080AC11E8D30EC204AFAF025A91F2</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1644</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>77.186386</td>\n",
       "      <td>12.939294</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>319.878021</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>34</th>\n",
       "      <td>34</td>\n",
       "      <td>2280153</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>A7F2CA5E11E8D72A15AB6E8B7925C4DE</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1645</td>\n",
       "      <td>False</td>\n",
       "      <td>True</td>\n",
       "      <td>...</td>\n",
       "      <td>75.924545</td>\n",
       "      <td>29.797392</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>298.001709</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>35</th>\n",
       "      <td>35</td>\n",
       "      <td>1980068</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>28C8789C11E8D23D208C959D736BCBEC</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1744</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>256.447388</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>36</th>\n",
       "      <td>36</td>\n",
       "      <td>1984733</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>2AD25E7211E8D24884247B9272A1FFC1</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1734</td>\n",
       "      <td>True</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>247.648254</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>37</th>\n",
       "      <td>37</td>\n",
       "      <td>2280461</td>\n",
       "      <td>CorruptedG</td>\n",
       "      <td>76561197992682099</td>\n",
       "      <td>A144304C11E8D72CE2866E8B24E68ECC</td>\n",
       "      <td>19</td>\n",
       "      <td>0</td>\n",
       "      <td>1645</td>\n",
       "      <td>False</td>\n",
       "      <td>False</td>\n",
       "      <td>...</td>\n",
       "      <td>75.176018</td>\n",
       "      <td>28.107063</td>\n",
       "      <td>False</td>\n",
       "      <td>1</td>\n",
       "      <td>320.899353</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "      <td>0.000000</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "<p>38 rows × 82 columns</p>\n",
       "</div>"
      ],
      "text/plain": [
       "    index       id        name             player  \\\n",
       "0       0  1810312  CorruptedG  76561197992682099   \n",
       "1       1  2280047  CorruptedG  76561197992682099   \n",
       "2       2  2310713  CorruptedG  76561197992682099   \n",
       "3       3  2433067  CorruptedG  76561197992682099   \n",
       "4       4  2279879  CorruptedG  76561197992682099   \n",
       "5       5  2448565  CorruptedG  76561197992682099   \n",
       "6       6  1982626  CorruptedG  76561197992682099   \n",
       "7       7  2351858  CorruptedG  76561197992682099   \n",
       "8       8  2311313  CorruptedG  76561197992682099   \n",
       "9       9  2452756  CorruptedG  76561197992682099   \n",
       "10     10  2414799  CorruptedG  76561197992682099   \n",
       "11     11  2414903  CorruptedG  76561197992682099   \n",
       "12     12  2433581  CorruptedG  76561197992682099   \n",
       "13     13  1498810  CorruptedG  76561197992682099   \n",
       "14     14  2414556  CorruptedG  76561197992682099   \n",
       "15     15  2312425  CorruptedG  76561197992682099   \n",
       "16     16  1498379  CorruptedG  76561197992682099   \n",
       "17     17  1498696  CorruptedG  76561197992682099   \n",
       "18     18  2356005  CorruptedG  76561197992682099   \n",
       "19     19  2280307  CorruptedG  76561197992682099   \n",
       "20     20  1926656  CorruptedG  76561197992682099   \n",
       "21     21  2356023  CorruptedG  76561197992682099   \n",
       "22     22  1984368  CorruptedG  76561197992682099   \n",
       "23     23  2280591  CorruptedG  76561197992682099   \n",
       "24     24  1498995  CorruptedG  76561197992682099   \n",
       "25     25  2020751  CorruptedG  76561197992682099   \n",
       "26     26  1985622  CorruptedG  76561197992682099   \n",
       "27     27  2311897  CorruptedG  76561197992682099   \n",
       "28     28  2520592  CorruptedG  76561197992682099   \n",
       "29     29  2326146  CorruptedG  76561197992682099   \n",
       "30     30  1982187  CorruptedG  76561197992682099   \n",
       "31     31  2415056  CorruptedG  76561197992682099   \n",
       "32     32  1498894  CorruptedG  76561197992682099   \n",
       "33     33  2074626  CorruptedG  76561197992682099   \n",
       "34     34  2280153  CorruptedG  76561197992682099   \n",
       "35     35  1980068  CorruptedG  76561197992682099   \n",
       "36     36  1984733  CorruptedG  76561197992682099   \n",
       "37     37  2280461  CorruptedG  76561197992682099   \n",
       "\n",
       "                                game  rank  division   mmr  is_orange    win  \\\n",
       "0   F42DF61811E8D0AFB57A4CBBE67B0A36    19         0  1712       True  False   \n",
       "1   99812C0A11E8D7296F926E8B15A8A97B    19         0  1645      False   True   \n",
       "2   68A4B98011E8D87D5CEB60A690E3C53A    19         0  1862      False   True   \n",
       "3   C794499E11E8DD3FD67A2DB278296A08    19         0  1932       True  False   \n",
       "4   A0EB8E3211E8D72888C06E8BA264708C    19         0  1645      False  False   \n",
       "5   16DDCCFA11E8CF444EA4C3A195DACF62    19         0  1561       True  False   \n",
       "6   0797E9E011E8D2433E145E963BE58303    19         0  1745       True  False   \n",
       "7   C709B9D811E8D89888C04BABA264708C    19         0  1684      False   True   \n",
       "8   93F3FF7A11E8D8810833C280918F6189    19         0  1862       True  False   \n",
       "9   4EF593E811E8D1BF003EF2957E06B85B    19         0  1706       True  False   \n",
       "10  F4A7E8C011E8DC7FBB623D9E327FF865    19         0  1703      False  False   \n",
       "11  6A4FEB6211E8DC815953989D60D4B411    19         0  1703      False   True   \n",
       "12  8A02699E11E8DD4447B6FBB5D3AEA3D3    19         0  1915       True  False   \n",
       "13  AE01BB0211E8C9AC87EF4A91777CAF14    19         0  1558      False   True   \n",
       "14  F56EB3FC11E8DC7EA5D4E2A31FFD83A0    19         0  1703      False   True   \n",
       "15  9FC51B8E11E8D888396DFB8A68244BC6    19         0  1868      False   True   \n",
       "16  AF3DA2A011E8C9ADD67A4A9178296A08    19         0  1558      False  False   \n",
       "17  C042DD8011E8C9AEF5C44A91BCD7229C    19         0  1558      False  False   \n",
       "18  8634AE0E11E8D9AB0FD7BD9D1B760370    19         0  1684       True   True   \n",
       "19  8FE6C37E11E8D72BF6B26E8B705CB0BC    19         0  1645      False  False   \n",
       "20  737245B211E8D1C01F441D93CE788AEF    19         0  1570       True  False   \n",
       "21  834B417011E8D9ACCB19AD8ADEE4DFF5    19         0  1684       True  False   \n",
       "22  1250A80011E8D247E993E3A0E1517F09    19         0  1734       True  False   \n",
       "23  AC25F35A11E8D72D7B426E8B6A3A4A60    19         0  1645      False   True   \n",
       "24  CFADFCC211E8C9AF66EB4A919DB5566D    19         0  1558      False  False   \n",
       "25  0FC4A22C11E8D29B1D9496B17D913D63    19         0  1625       True   True   \n",
       "26  C0DF35BA11E8D249E6399B9BDD4705C1    19         0  1734      False  False   \n",
       "27  C2475BE411E8D8847EA2F3941DE3A62F    19         0  1868       True   True   \n",
       "28  710CEACC11E8DFF6CD02CDA535F4529D    19         0  1737      False   True   \n",
       "29  9D45BB9E11E8D8F07ABAEE9FE3DA7507    19         0  1908      False   True   \n",
       "30  E332DB4211E8D241F5C4B092BCD7229C    19         0  1745       True  False   \n",
       "31  4E0F1DC811E8DC828ACD30A07597FDD4    19         0  1703      False   True   \n",
       "32  834EE2F011E8C9ABD5514A91DFFAAFCB    19         0  1558      False   True   \n",
       "33  C0F080AC11E8D30EC204AFAF025A91F2    19         0  1644       True  False   \n",
       "34  A7F2CA5E11E8D72A15AB6E8B7925C4DE    19         0  1645      False   True   \n",
       "35  28C8789C11E8D23D208C959D736BCBEC    19         0  1744       True  False   \n",
       "36  2AD25E7211E8D24884247B9272A1FFC1    19         0  1734       True  False   \n",
       "37  A144304C11E8D72CE2866E8B24E68ECC    19         0  1645      False  False   \n",
       "\n",
       "            ...           time_closest_to_team_center  \\\n",
       "0           ...                              0.000000   \n",
       "1           ...                             90.470673   \n",
       "2           ...                              0.000000   \n",
       "3           ...                              0.000000   \n",
       "4           ...                             71.419067   \n",
       "5           ...                             73.591843   \n",
       "6           ...                              0.000000   \n",
       "7           ...                             59.694107   \n",
       "8           ...                              0.000000   \n",
       "9           ...                             64.702675   \n",
       "10          ...                            136.537918   \n",
       "11          ...                             78.854057   \n",
       "12          ...                              0.000000   \n",
       "13          ...                             76.840515   \n",
       "14          ...                             80.277771   \n",
       "15          ...                              0.000000   \n",
       "16          ...                             81.425682   \n",
       "17          ...                             72.720779   \n",
       "18          ...                             52.461136   \n",
       "19          ...                             79.470146   \n",
       "20          ...                             76.285492   \n",
       "21          ...                             95.315987   \n",
       "22          ...                              0.000000   \n",
       "23          ...                             72.441048   \n",
       "24          ...                             74.547325   \n",
       "25          ...                             84.157669   \n",
       "26          ...                              0.000000   \n",
       "27          ...                              0.000000   \n",
       "28          ...                             64.296173   \n",
       "29          ...                              0.000000   \n",
       "30          ...                              0.000000   \n",
       "31          ...                             83.874092   \n",
       "32          ...                             71.580940   \n",
       "33          ...                             77.186386   \n",
       "34          ...                             75.924545   \n",
       "35          ...                              0.000000   \n",
       "36          ...                              0.000000   \n",
       "37          ...                             75.176018   \n",
       "\n",
       "    time_furthest_from_team_center  is_bot  first_frame_in_game  time_in_game  \\\n",
       "0                         0.000000   False                    1    316.817078   \n",
       "1                        23.071768   False                    1    320.870453   \n",
       "2                         0.000000   False                    1    309.791565   \n",
       "3                         0.000000   False                    1    239.715698   \n",
       "4                        15.992106   False                    1    304.897400   \n",
       "5                        12.885868   False                    1    317.122864   \n",
       "6                         0.000000   False                    1    267.241119   \n",
       "7                        14.634933   False                    1    322.558838   \n",
       "8                         0.000000   False                    1    279.595490   \n",
       "9                        14.938508   False                    1    307.896698   \n",
       "10                       39.434074   False                    1    559.563354   \n",
       "11                       19.491844   False                    1    309.838928   \n",
       "12                        0.000000   False                    1    191.794312   \n",
       "13                        7.734396   False                    1    320.613831   \n",
       "14                       20.290960   False                    1    305.823883   \n",
       "15                        0.000000   False                    1    339.367737   \n",
       "16                       14.498387   False                    1    311.646606   \n",
       "17                        7.823503   False                    1    311.937500   \n",
       "18                       21.522015   False                    1    308.718506   \n",
       "19                       44.426254   False                    1    358.820129   \n",
       "20                       25.516151   False                    1    316.683105   \n",
       "21                       22.648022   False                    1    338.979065   \n",
       "22                        0.000000   False                    1    323.215607   \n",
       "23                       20.536903   False                    1    319.386444   \n",
       "24                       25.207628   False                    1    314.491699   \n",
       "25                       25.863363   False                    1    315.624603   \n",
       "26                        0.000000   False                    1    362.147217   \n",
       "27                        0.000000   False                    1    416.602844   \n",
       "28                       13.021171   False                    1    315.299683   \n",
       "29                        0.000000   False                    1    325.902771   \n",
       "30                        0.000000   False                    1    345.010986   \n",
       "31                       20.201830   False                    1    357.592102   \n",
       "32                       17.121521   False                    1    315.177673   \n",
       "33                       12.939294   False                    1    319.878021   \n",
       "34                       29.797392   False                    1    298.001709   \n",
       "35                        0.000000   False                    1    256.447388   \n",
       "36                        0.000000   False                    1    247.648254   \n",
       "37                       28.107063   False                    1    320.899353   \n",
       "\n",
       "    time_in_front_of_center_of_mass  time_behind_center_of_mass  \\\n",
       "0                          0.000000                    0.000000   \n",
       "1                          0.000000                    0.000000   \n",
       "2                          0.000000                    0.000000   \n",
       "3                          0.000000                    0.000000   \n",
       "4                          0.000000                    0.000000   \n",
       "5                        129.079178                  188.043671   \n",
       "6                          0.000000                    0.000000   \n",
       "7                        142.799973                  179.758850   \n",
       "8                          0.000000                    0.000000   \n",
       "9                        156.848862                  151.047836   \n",
       "10                       206.932556                  352.593109   \n",
       "11                       148.863434                  160.975494   \n",
       "12                         0.000000                    0.000000   \n",
       "13                         0.000000                    0.000000   \n",
       "14                       110.405708                  195.418182   \n",
       "15                         0.000000                    0.000000   \n",
       "16                         0.000000                    0.000000   \n",
       "17                         0.000000                    0.000000   \n",
       "18                       141.135162                  167.583359   \n",
       "19                         0.000000                    0.000000   \n",
       "20                         0.000000                    0.000000   \n",
       "21                       155.768936                  183.100800   \n",
       "22                         0.000000                    0.000000   \n",
       "23                         0.000000                    0.000000   \n",
       "24                         0.000000                    0.000000   \n",
       "25                         0.000000                    0.000000   \n",
       "26                         0.000000                    0.000000   \n",
       "27                         0.000000                    0.000000   \n",
       "28                       155.757812                  159.541870   \n",
       "29                         0.000000                    0.000000   \n",
       "30                         0.000000                    0.000000   \n",
       "31                       187.077820                  170.514282   \n",
       "32                         0.000000                    0.000000   \n",
       "33                         0.000000                    0.000000   \n",
       "34                         0.000000                    0.000000   \n",
       "35                         0.000000                    0.000000   \n",
       "36                         0.000000                    0.000000   \n",
       "37                         0.000000                    0.000000   \n",
       "\n",
       "    time_most_forward_player  time_most_back_player  time_between_players  \n",
       "0                   0.000000               0.000000              0.000000  \n",
       "1                   0.000000               0.000000              0.000000  \n",
       "2                   0.000000               0.000000              0.000000  \n",
       "3                   0.000000               0.000000              0.000000  \n",
       "4                   0.000000               0.000000              0.000000  \n",
       "5                  93.191521             136.420151             87.511192  \n",
       "6                   0.000000               0.000000              0.000000  \n",
       "7                 103.126450             122.617470             96.814919  \n",
       "8                   0.000000               0.000000              0.000000  \n",
       "9                 112.929527              91.868462            103.098709  \n",
       "10                122.960953             259.256866            177.345535  \n",
       "11                 99.109039             116.117851             94.612038  \n",
       "12                  0.000000               0.000000              0.000000  \n",
       "13                  0.000000               0.000000              0.000000  \n",
       "14                 63.209217             137.904694            104.709976  \n",
       "15                  0.000000               0.000000              0.000000  \n",
       "16                  0.000000               0.000000              0.000000  \n",
       "17                  0.000000               0.000000              0.000000  \n",
       "18                 99.395424             118.196808             91.126274  \n",
       "19                  0.000000               0.000000              0.000000  \n",
       "20                  0.000000               0.000000              0.000000  \n",
       "21                 97.884315             127.128174            113.966576  \n",
       "22                  0.000000               0.000000              0.000000  \n",
       "23                  0.000000               0.000000              0.000000  \n",
       "24                  0.000000               0.000000              0.000000  \n",
       "25                  0.000000               0.000000              0.000000  \n",
       "26                  0.000000               0.000000              0.000000  \n",
       "27                  0.000000               0.000000              0.000000  \n",
       "28                108.971848             109.194321             97.133514  \n",
       "29                  0.000000               0.000000              0.000000  \n",
       "30                  0.000000               0.000000              0.000000  \n",
       "31                121.785156             124.780144            111.026802  \n",
       "32                  0.000000               0.000000              0.000000  \n",
       "33                  0.000000               0.000000              0.000000  \n",
       "34                  0.000000               0.000000              0.000000  \n",
       "35                  0.000000               0.000000              0.000000  \n",
       "36                  0.000000               0.000000              0.000000  \n",
       "37                  0.000000               0.000000              0.000000  \n",
       "\n",
       "[38 rows x 82 columns]"
      ]
     },
     "execution_count": 625,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# smurf_data = concat_data([get_playergames_by_id(smurfs[2])])\n",
    "# smurf_data = concat_data([get_playergames_by_id('76561198060924319')])\n",
    "# smurf_data = concat_data([get_playergames_by_id('76561198055442516')])\n",
    "smurf_data = concat_data([get_playergames_by_id('76561197992682099')])\n",
    "\n",
    "smurf_data"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": 626,
   "metadata": {},
   "outputs": [],
   "source": [
    "point = process_input_data(smurf_data, maxs, mins)[0].values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 627,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([18, 18, 18, 17, 18, 18, 17, 19, 17, 15, 19, 18, 18, 19, 19, 18, 19,\n",
       "       19, 19, 19, 19, 19, 15, 19, 18, 17, 18, 18, 17, 18, 18, 18, 19, 18,\n",
       "       19, 17, 18, 19])"
      ]
     },
     "execution_count": 627,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "result = pd.DataFrame(index=list(range(len(point))))\n",
    "for rank, m in models.items():\n",
    "    result = result.merge(pd.Series(100 * m(point.astype(float)).detach().cpu().numpy().reshape(len(smurf_data)), name=str(rank)).to_frame(), left_index=True, right_index=True)\n",
    "result = (result > 50).apply(lambda x: int(x.idxmin()), axis=1).values\n",
    "result"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 628,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "18.05263157894737\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "18.0"
      ]
     },
     "execution_count": 628,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "print(result.mean())\n",
    "round(result.mean())"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# PCA"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 591,
   "metadata": {},
   "outputs": [],
   "source": [
    "k = 3"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 592,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(49814, 67)"
      ]
     },
     "execution_count": 592,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "input.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 593,
   "metadata": {},
   "outputs": [],
   "source": [
    "def PCA(data, offset=0, k=2):\n",
    "    # preprocess the data\n",
    "    X = torch.from_numpy(data.values.astype(float))\n",
    "    X_mean = torch.mean(X,0)\n",
    "    X = X - X_mean.expand_as(X)\n",
    "\n",
    "    # svd\n",
    "    U,S,V = torch.svd(torch.t(X))\n",
    "    print(S.round(), S.max(), S.argmax())\n",
    "    print(U.shape, S.shape, V.shape)\n",
    "    print (U[0, :50].argmax())\n",
    "    return torch.mm(X,U[:,offset:offset + k])\n",
    "def PCA_col(data, columns=[18, 19]):\n",
    "    # preprocess the data\n",
    "    X = torch.from_numpy(data.values.astype(float))\n",
    "    X_mean = torch.mean(X,0)\n",
    "    X = X - X_mean.expand_as(X)\n",
    "\n",
    "    # svd\n",
    "    U,S,V = torch.svd(torch.t(X))\n",
    "#     print(U.shape)\n",
    "    return torch.mm(X,U[:,columns])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 601,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "Text(0, 0.5, 'average_speed')"
      ]
     },
     "execution_count": 601,
     "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": [
    "mi = U[:, 0].numpy().argsort() - 1\n",
    "important = input.columns[mi]\n",
    "for i in [10, 19]:\n",
    "    plt.scatter(input[ranks == i][important[0]], input[ranks == i][important[1]], label=str(i))\n",
    "plt.legend()\n",
    "plt.xlabel(important[0])\n",
    "plt.ylabel(important[1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 595,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "tensor([31., 21., 16., 15., 13., 10., 10.,  9.,  8.,  8.,  7.,  7.,  6.,  6.,\n",
      "         5.,  5.,  5.,  5.,  4.,  4.,  4.,  4.,  4.,  4.,  3.,  3.,  3.,  3.,\n",
      "         2.,  2.,  2.,  2.,  2.,  2.,  2.,  2.,  1.,  1.,  1.,  1.,  1.,  1.,\n",
      "         1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  1.,  0.,\n",
      "         0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.,  0.],\n",
      "       dtype=torch.float64) tensor(30.5607, dtype=torch.float64) tensor(0)\n",
      "torch.Size([67, 67]) torch.Size([67]) torch.Size([9947, 67])\n",
      "tensor(38)\n"
     ]
    },
    {
     "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": [
    "ranks = output.reshape(len(output))\n",
    "pred = (ranks == 10) | (ranks == 19)\n",
    "for offset in range(68):\n",
    "#     offset = 2\n",
    "    k = 2\n",
    "    X_PCA = PCA(input[pred], offset=offset, k=k)\n",
    "    plt.figure()\n",
    "    y = torch.from_numpy(ranks[pred])\n",
    "    n = 100\n",
    "    for i in [10, 19]:#range(10, 20):\n",
    "        plt.scatter(X_PCA[y == i, 0][:n], X_PCA[y == i, 1][:n], label=str(i), alpha=0.5)\n",
    "\n",
    "    plt.legend()\n",
    "    plt.title('PCA of ' + \", \".join([str(s) for s in list(range(offset, offset + k))]))\n",
    "    plt.show()\n",
    "    break"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 596,
   "metadata": {},
   "outputs": [
    {
     "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": [
    "ranks = output.reshape(len(output))\n",
    "pred = (ranks == 10) | (ranks == 19)\n",
    "columns=[4, 7]\n",
    "k = 2\n",
    "X_PCA = PCA_col(input[pred], columns=columns)\n",
    "plt.figure()\n",
    "y = torch.from_numpy(ranks[pred])\n",
    "n = 200\n",
    "for i in [10, 19]:#range(10, 20):\n",
    "    plt.scatter(X_PCA[y == i, 0][:n], X_PCA[y == i, 1][:n], label=str(i), alpha=0.3)\n",
    "\n",
    "plt.legend()\n",
    "plt.title('PCA of ' + \", \".join([str(s) for s in columns]))\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 597,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "4980"
      ]
     },
     "execution_count": 597,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(ranks == 10).sum()\n",
    "(ranks == 19).sum()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 459,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "tensor([], dtype=torch.float64)"
      ]
     },
     "execution_count": 459,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_PCA[, :]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 456,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(49814,)"
      ]
     },
     "execution_count": 456,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "y.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 461,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "torch.Size([49814])"
      ]
     },
     "execution_count": 461,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "(X_PCA[:, 0] == 0).shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "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": "3.6.6"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}