Car_model.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-12T10:31:13.733534Z",
     "start_time": "2021-02-12T10:31:13.727190Z"
    }
   },
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "\n",
    "import xgboost as xgb\n",
    "import lightgbm as lgb\n",
    "\n",
    "from sklearn.dummy import DummyRegressor\n",
    "from sklearn.tree import DecisionTreeRegressor\n",
    "from sklearn.ensemble import RandomForestRegressor\n",
    "\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "\n",
    "from sklearn.metrics import mean_absolute_error as mae\n",
    "from sklearn.model_selection import cross_val_score, KFold\n",
    "\n",
    "import eli5\n",
    "from eli5.sklearn import PermutationImportance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-12T10:31:12.661144Z",
     "start_time": "2021-02-12T10:31:12.657888Z"
    }
   },
   "outputs": [],
   "source": [
    "pd.set_option('display.max_columns', 250)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-12T10:31:13.724667Z",
     "start_time": "2021-02-12T10:31:12.663873Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(201357, 101)"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df = pd.read_pickle('data/car_clean.pickle')\n",
    "df.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-12T10:31:17.476702Z",
     "start_time": "2021-02-12T10:31:17.168132Z"
    }
   },
   "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": [
    "sns.distplot(np.log2(df['Cena']));"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-12T10:31:22.888964Z",
     "start_time": "2021-02-12T10:31:22.882522Z"
    }
   },
   "outputs": [],
   "source": [
    "def run_model(model, df, feats):\n",
    "    X = df[ feats ].values\n",
    "    y = df['Cena'].values\n",
    "    \n",
    "    scores = []\n",
    "    \n",
    "    folds = KFold(n_splits=3, shuffle=True, random_state=0)\n",
    "    for train_index, test_index in folds.split(X, y):\n",
    "        X_train, X_test = X[train_index], X[test_index]\n",
    "        y_train, y_test = y[train_index], y[test_index]\n",
    "        \n",
    "        y_train = np.log2( y_train )\n",
    "        model.fit(X_train, y_train)\n",
    "        \n",
    "        y_pred = model.predict(X_test)\n",
    "        y_pred = np.exp2( y_pred )\n",
    "        scores.append(mae(y_test, y_pred))\n",
    "        \n",
    "#     scores = cross_val_score(model, X, y, cv = 3, scoring = \"neg_mean_absolute_error\" )\n",
    "    return np.mean(scores), np.std(scores)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-12T10:31:25.355728Z",
     "start_time": "2021-02-12T10:31:25.334905Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array(['Model pojazdu', 'Wersja', 'Generacja', 'Opis'], dtype=object)"
      ]
     },
     "execution_count": 7,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# cat_feats = df.select_dtypes(np.object).columns.values\n",
    "# cat_feats"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-09T11:11:30.901516Z",
     "start_time": "2021-02-09T11:11:25.419517Z"
    }
   },
   "outputs": [],
   "source": [
    "# SUFF = '__cat'\n",
    "# for feat in cat_feats[:-1]:\n",
    "#     size = len(df[feat].unique())\n",
    "#     if size >= 10:\n",
    "#         df[feat + SUFF] = df[feat].factorize()[0]\n",
    "#     else:\n",
    "#         df = pd.concat([df,pd.get_dummies(df[feat], prefix=feat)], axis=1)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-12T10:31:41.743702Z",
     "start_time": "2021-02-12T10:31:41.741049Z"
    }
   },
   "outputs": [],
   "source": [
    "black_list = ['Cena', 'Model pojazdu__cat']"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-12T10:31:42.324470Z",
     "start_time": "2021-02-12T10:31:42.233177Z"
    }
   },
   "outputs": [],
   "source": [
    "basic_feats = df.select_dtypes(np.number).columns.values\n",
    "basic_feats = [x for x in basic_feats if x not in black_list]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-12T10:32:03.233445Z",
     "start_time": "2021-02-12T10:32:03.038484Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "819611.8000000279"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "up = np.percentile(df['Cena'],99.9)\n",
    "df1 = df.copy()\n",
    "df1 = df1[df1['Cena'] < up]\n",
    "up"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-12T10:32:06.846311Z",
     "start_time": "2021-02-12T10:32:06.343306Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(43529.1327031434, 20.461128572061934)"
      ]
     },
     "execution_count": 12,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = DummyRegressor(strategy='median')\n",
    "run_model(model, df1, basic_feats)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-09T11:50:07.925334Z",
     "start_time": "2021-02-09T11:50:02.054874Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(16842.24233725219, 193.6602152548799)"
      ]
     },
     "execution_count": 53,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = DecisionTreeRegressor(max_depth=5, random_state=0)\n",
    "run_model(model, df1, basic_feats)#19100"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-09T11:52:57.061354Z",
     "start_time": "2021-02-09T11:50:15.219566Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(15994.653771752424, 308.0445023202748)"
      ]
     },
     "execution_count": 54,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = RandomForestRegressor(max_depth=5, n_estimators = 50, random_state=0)\n",
    "run_model(model, df1, basic_feats)#18221"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-09T11:53:34.771188Z",
     "start_time": "2021-02-09T11:52:57.063655Z"
    }
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(9750.943700518872, 99.59635044273006)"
      ]
     },
     "execution_count": 55,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = xgb.XGBRegressor(max_depth = 5, n_estimators = 50, random_state=0)\n",
    "run_model(model, df1, basic_feats)#13682"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-12T10:32:41.924610Z",
     "start_time": "2021-02-12T10:32:39.015776Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[LightGBM] [Warning] Accuracy may be bad since you didn't explicitly set num_leaves OR 2^max_depth > num_leaves. (num_leaves=31).\n",
      "[LightGBM] [Warning] Accuracy may be bad since you didn't explicitly set num_leaves OR 2^max_depth > num_leaves. (num_leaves=31).\n",
      "[LightGBM] [Warning] Accuracy may be bad since you didn't explicitly set num_leaves OR 2^max_depth > num_leaves. (num_leaves=31).\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "(10608.862977760313, 118.87742245083223)"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "# perhaps the hyperparameters can be further optimized, but the time-to-result ratio is optimal here\n",
    "\n",
    "lgbm_parms = {\n",
    "    'max_depth': 5,\n",
    "    'n_estimators': 50,\n",
    "    'random_state': 0\n",
    "}\n",
    "model = lgb.LGBMRegressor(**lgbm_parms)\n",
    "run_model(model, df1, basic_feats)\n",
    "# score (6972.534592681524, 51.97959183884229)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-12T11:28:30.299213Z",
     "start_time": "2021-02-12T11:28:30.069724Z"
    }
   },
   "outputs": [],
   "source": [
    "model = lgb.Booster(model_file = './data/lgb_model_easy.txt')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-12T11:29:06.915957Z",
     "start_time": "2021-02-12T11:28:31.261795Z"
    }
   },
   "outputs": [],
   "source": [
    "y_pred = model.predict(df1[basic_feats])\n",
    "y_pred = np.exp2( y_pred )"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {
    "ExecuteTime": {
     "end_time": "2021-02-12T11:29:06.927692Z",
     "start_time": "2021-02-12T11:29:06.919422Z"
    }
   },
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "3541.779498565464\n"
     ]
    }
   ],
   "source": [
    "print(mae(df1['Cena'], y_pred))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "hide_input": false,
  "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": "3.7.2"
  },
  "toc": {
   "base_numbering": 1,
   "nav_menu": {},
   "number_sections": true,
   "sideBar": true,
   "skip_h1_title": false,
   "title_cell": "Table of Contents",
   "title_sidebar": "Contents",
   "toc_cell": false,
   "toc_position": {},
   "toc_section_display": true,
   "toc_window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}