Observer.2.1.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 193,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "import sklearn.cluster as cluster\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.preprocessing import StandardScaler as SC\n",
    "from sklearn.neighbors import KNeighborsRegressor as KNN\n",
    "from sklearn.preprocessing import LabelEncoder\n",
    "\n",
    "big_data = pd.read_csv('airbnb_coded.2.0.csv')\n",
    "sc = SC()\n",
    "sc_y = SC()\n",
    "le = LabelEncoder()\n",
    "knn = KNN(n_neighbors=65, weights='distance')\n",
    "\n",
    "pd.options.mode.chained_assignment = None  # default='warn'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 215,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "70\n",
      "9000.0\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.compose import ColumnTransformer\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "\n",
    "ct = ColumnTransformer(transformers=[('encoder', OneHotEncoder(), [-1])], remainder='passthrough')\n",
    "var = 'bathrooms'\n",
    "var2 = 'accommodates'\n",
    "var3 = 'cleaning_fee'\n",
    "var4 = 'bedrooms'\n",
    "var5 = 'beds'\n",
    "var6 = 'minimum_nights_x'\n",
    "var7 = 'maximum_nights'\n",
    "var8 = 'number_of_reviews_y'\n",
    "var9 = 'listing_class'\n",
    "var10 = 'price_x'\n",
    "var11 = 'bed_code'\n",
    "var12 = 'cancellation_policy_mapped'\n",
    "var13 = 'room_code'\n",
    "var14 = 'superhost'\n",
    "var15 = 'instant_book'\n",
    "var16 = 'review_scores_location'\n",
    "var17 = 'review_scores_rating'\n",
    "var18 = 'host_total_listings_count'\n",
    "var19 = 'availability_90'\n",
    "var20 = 'room_code'\n",
    "var21 = 'extra_people'\n",
    "var22 = 'property_type'\n",
    "\n",
    "\n",
    "data = big_data[[var18, var2, var3, var22, var10]]\n",
    "\n",
    "\n",
    "# data[var] = data[var].fillna(data[var].median())\n",
    "data[var2] = data[var2].fillna(data[var2].median())\n",
    "data[var3] = data[var3].fillna(data[var3].median())\n",
    "# data[var4] = data[var4].fillna(data[var4].median())\n",
    "# data[var5] = data[var5].fillna(data[var5].median())\n",
    "\n",
    "# data[var6][17990] = data[var6].median()\n",
    "# data[var6][6215] = data[var6].median()\n",
    "# data[var6][11299] = data[var6].median()\n",
    "# data[var6][12737] = data[var6].median()\n",
    "# data[var6][14598] = data[var6].median()\n",
    "# data[var7][872] = 8000\n",
    "# data[var7][2372] = 8000\n",
    "# data[var7][4391] = 8000\n",
    "# data[var7][17952] = 8000\n",
    "# data[var7][17990] = data[var7].median()\n",
    "# data[var7][6351] = data[var7].median()\n",
    "# data[var7][8341] = data[var7].median()\n",
    "\n",
    "\n",
    "# data[var6] = data[var6].fillna(data[var6].median())\n",
    "# data[var7] = data[var7].fillna(data[var7].median())\n",
    "# data[var8] = data[var8].fillna(data[var8].median())\n",
    "data[var10] = data[var10].fillna(data[var10].median())\n",
    "# data[var11] = data[var11].fillna(data[var11].median())\n",
    "# data[var12] = data[var12].fillna(data[var12].median())\n",
    "# data[var13] = data[var13].fillna(data[var13].median())\n",
    "# data[var14] = data[var14].fillna(data[var14].median())\n",
    "# data[var15] = data[var15].fillna(data[var15].median())\n",
    "# data[var16] = data[var16].fillna(data[var16].median())\n",
    "# data[var17] = data[var17].fillna(data[var17].median())\n",
    "data[var18] = data[var18].fillna(data[var18].median())\n",
    "# data[var19] = data[var19].fillna(data[var19].median())\n",
    "# data[var20] = data[var20].fillna(data[var20].median())\n",
    "# data[var21] = data[var21].fillna(data[var21].median())\n",
    "data['property_type'] = data['property_type'].fillna('Apartment')\n",
    "\n",
    "\n",
    "# data[var6][17990] = data[var6].median()\n",
    "\n",
    "X = data.iloc[:, :-1].values\n",
    "y = data.iloc[:, -1].values\n",
    "X = ct.fit_transform(X)\n",
    "X = X.todense()\n",
    "X = np.array(X)\n",
    "# X = ct.fit_transform(X)\n",
    "# # X = X.todense()\n",
    "# X = np.array(X)\n",
    "# X = ct.fit_transform(X)\n",
    "# # X = X.todense()\n",
    "# X = np.array(X)\n",
    "# X = ct.fit_transform(X)\n",
    "# # X = X.todense()\n",
    "# X = np.array(X)\n",
    "# # ct = ColumnTransformer(transformers=[('encoder', OneHotEncoder(), [-4])], remainder='passthrough')\n",
    "# X = ct.fit_transform(X)\n",
    "# # X = X.todense()\n",
    "# X = np.array(X)\n",
    "# ct = ColumnTransformer(transformers=[('encoder', OneHotEncoder(), [-5])], remainder='passthrough')\n",
    "# X = ct.fit_transform(X)\n",
    "# X = X.todense()\n",
    "# X = np.array(X)\n",
    "# ct = ColumnTransformer(transformers=[('encoder', OneHotEncoder(), [-6])], remainder='passthrough')\n",
    "# X = ct.fit_transform(X)\n",
    "# # X = X.todense()\n",
    "# X = np.array(X)\n",
    "\n",
    "# plt.scatter(data[var6].values, data[var7].values, s=100, c='red', alpha=0.5)\n",
    "# plt.show()\n",
    "\n",
    "# X[:,-8:] = sc.fit_transform(X[:,-8:])\n",
    "# X = sc.fit_transform(X)\n",
    "counter=0\n",
    "for i in range(len(y)):\n",
    "    if y[i]>=500:\n",
    "        counter+=1\n",
    "print(counter)\n",
    "print(max(y.T))\n",
    "# X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.1)\n",
    "\n",
    "# data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 195,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.collections.PathCollection at 0x7fd118c83310>"
      ]
     },
     "execution_count": 195,
     "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(data[var22].values, data[var10].values, alpha=0.2)\n"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 196,
   "metadata": {},
   "outputs": [],
   "source": [
    "# from hdbscan import HDBSCAN as HDB\n",
    "# hdb = HDB(min_cluster_size=10, min_samples=16)\n",
    "# price_cluster = hdb.fit_predict(X)\n",
    "# price_cluster\n",
    "# data['price_class'] = price_cluster\n",
    "# big_data['price_class'] = price_cluster\n",
    "# data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 197,
   "metadata": {},
   "outputs": [],
   "source": [
    "# counter = 0\n",
    "# for index in price_cluster:\n",
    "#     if index == -1:\n",
    "#         counter +=1\n",
    "# counter"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 198,
   "metadata": {},
   "outputs": [],
   "source": [
    "# clusterer = cluster.KMeans(n_clusters=4)\n",
    "# price_cluster = clusterer.fit_predict(X)\n",
    "# price_cluster\n",
    "# data['price_class'] = price_cluster\n",
    "# big_data['price_class'] = price_cluster\n",
    "# data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 199,
   "metadata": {},
   "outputs": [],
   "source": [
    "# plt.scatter(data[var3].values, data['price_class'].values, alpha=0.1, c='red')\n",
    "# plt.scatter(data[var10].values, data['price_class'].values, alpha=0.1, c='blue')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 200,
   "metadata": {},
   "outputs": [],
   "source": [
    "big_data.to_csv('airbnb_coded.3.0.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 155,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import seaborn as sns\n",
    "import sklearn.cluster as cluster\n",
    "import numpy as np\n",
    "import matplotlib.pyplot as plt\n",
    "from sklearn.preprocessing import StandardScaler as SC\n",
    "from sklearn.neighbors import KNeighborsRegressor as KNN\n",
    "\n",
    "big_data = pd.read_csv('airbnb_coded.3.0.csv')\n",
    "sc = SC()\n",
    "sc_y = SC()\n",
    "knn = KNN(n_neighbors=65, weights='distance')\n",
    "\n",
    "pd.options.mode.chained_assignment = None  # default='warn'"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 156,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[[  1.   0.   0. ...  97.   0.   0.]\n",
      " [  1.   0.   0. ...  75.   0.   0.]\n",
      " [  0.   0.   0. ...  80.  10.  83.]\n",
      " ...\n",
      " [  0.   0.   0. ...  99.  19.   6.]\n",
      " [  0.   0.   0. ...  90.  10.   0.]\n",
      " [  1.   0.   0. ... 100.   0.   0.]]\n",
      "25\n"
     ]
    }
   ],
   "source": [
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.compose import ColumnTransformer\n",
    "from sklearn.preprocessing import OneHotEncoder\n",
    "\n",
    "ct = ColumnTransformer(transformers=[('encoder', OneHotEncoder(), [-1])], remainder='passthrough')\n",
    "var = 'bathrooms'\n",
    "var2 = 'accommodates'\n",
    "var3 = 'cleaning_fee'\n",
    "var4 = 'bedrooms'\n",
    "var5 = 'beds'\n",
    "var6 = 'minimum_nights_x'\n",
    "var7 = 'maximum_nights'\n",
    "var8 = 'number_of_reviews_y'\n",
    "var9 = 'listing_class'\n",
    "var10 = 'price_x'\n",
    "var11 = 'bed_code'\n",
    "var12 = 'cancellation_policy_mapped'\n",
    "var13 = 'room_code'\n",
    "var14 = 'superhost'\n",
    "var15 = 'instant_book'\n",
    "var16 = 'review_scores_location'\n",
    "var17 = 'review_scores_rating'\n",
    "var18 = 'host_total_listings_count'\n",
    "var19 = 'availability_90'\n",
    "var20 = 'room_code'\n",
    "var21 = 'extra_people'\n",
    "var22 = 'price_class'\n",
    "\n",
    "\n",
    "data = big_data[[var18, var2, var3, var, var6, var8,  var17, var21, var19, var9, var12, var20, var22, var14, var10]]\n",
    "\n",
    "data[var] = data[var].fillna(data[var].median())\n",
    "data[var2] = data[var2].fillna(data[var2].median())\n",
    "data[var3] = data[var3].fillna(data[var3].median())\n",
    "# data[var4] = data[var4].fillna(data[var4].median())\n",
    "# data[var5] = data[var5].fillna(data[var5].median())\n",
    "\n",
    "data[var6][17990] = data[var6].median()\n",
    "data[var6][6215] = data[var6].median()\n",
    "data[var6][11299] = data[var6].median()\n",
    "data[var6][12737] = data[var6].median()\n",
    "data[var6][14598] = data[var6].median()\n",
    "# data[var7][872] = 8000\n",
    "# data[var7][2372] = 8000\n",
    "# data[var7][4391] = 8000\n",
    "# data[var7][17952] = 8000\n",
    "# data[var7][17990] = data[var7].median()\n",
    "# data[var7][6351] = data[var7].median()\n",
    "# data[var7][8341] = data[var7].median()\n",
    "\n",
    "\n",
    "data[var6] = data[var6].fillna(data[var6].median())\n",
    "# data[var7] = data[var7].fillna(data[var7].median())\n",
    "data[var8] = data[var8].fillna(data[var8].median())\n",
    "data[var10] = data[var10].fillna(data[var10].median())\n",
    "# data[var11] = data[var11].fillna(data[var11].median())\n",
    "data[var12] = data[var12].fillna(data[var12].median())\n",
    "# data[var13] = data[var13].fillna(data[var13].median())\n",
    "data[var14] = data[var14].fillna(data[var14].median())\n",
    "# data[var15] = data[var15].fillna(data[var15].median())\n",
    "# data[var16] = data[var16].fillna(data[var16].median())\n",
    "data[var17] = data[var17].fillna(data[var17].median())\n",
    "data[var18] = data[var18].fillna(data[var18].median())\n",
    "data[var19] = data[var19].fillna(data[var19].median())\n",
    "data[var20] = data[var20].fillna(data[var20].median())\n",
    "data[var21] = data[var21].fillna(data[var21].median())\n",
    "\n",
    "\n",
    "data[var6][17990] = data[var6].median()\n",
    "\n",
    "data = data.drop(data.index[17990])\n",
    "data = data.drop(data.index[14598])\n",
    "\n",
    "\n",
    "\n",
    "X = data.iloc[:, :-1].values\n",
    "y = data.iloc[:, -1].values\n",
    "price_calss = data.iloc[:, 0].values\n",
    "X = ct.fit_transform(X)\n",
    "# X = X.todense()\n",
    "X = np.array(X)\n",
    "X = ct.fit_transform(X)\n",
    "# X = X.todense()\n",
    "X = np.array(X)\n",
    "X = ct.fit_transform(X)\n",
    "# X = X.todense()\n",
    "X = np.array(X)\n",
    "X = ct.fit_transform(X)\n",
    "# X = X.todense()\n",
    "X = np.array(X)\n",
    "# ct = ColumnTransformer(transformers=[('encoder', OneHotEncoder(), [-4])], remainder='passthrough')\n",
    "X = ct.fit_transform(X)\n",
    "# X = X.todense()\n",
    "X = np.array(X)\n",
    "# ct = ColumnTransformer(transformers=[('encoder', OneHotEncoder(), [-5])], remainder='passthrough')\n",
    "# X = ct.fit_transform(X)\n",
    "# X = X.todense()\n",
    "# X = np.array(X)\n",
    "# ct = ColumnTransformer(transformers=[('encoder', OneHotEncoder(), [-6])], remainder='passthrough')\n",
    "# X = ct.fit_transform(X)\n",
    "# # X = X.todense()\n",
    "# X = np.array(X)\n",
    "\n",
    "# plt.scatter(data[var6].values, data[var7].values, s=100, c='red', alpha=0.5)\n",
    "# plt.show()\n",
    "\n",
    "print(X)\n",
    "# X[:,-8:] = sc.fit_transform(X[:,-8:])\n",
    "# X = sc.fit_transform(X)\n",
    "print(len(X[0]))\n",
    "# X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.1)\n",
    "\n",
    "# data.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 157,
   "metadata": {},
   "outputs": [],
   "source": [
    "# X_cat = data[[var18, var2, var3]].values\n",
    "# y_cat = data[[var22]].values\n",
    "# X_train, X_test, y_train, y_test = train_test_split(X_cat, y_cat, test_size = 0.3)\n",
    "# from sklearn.tree import DecisionTreeClassifier\n",
    "# classifier = DecisionTreeClassifier(max_depth=2)\n",
    "# classifier = classifier.fit(X_train, y_train)\n",
    "# print(classifier.score(X_train, y_train))\n",
    "# classifier.score(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 158,
   "metadata": {},
   "outputs": [],
   "source": [
    "# X_cat = data[[var18, var2, var3]].values\n",
    "# y_cat = data[[var22]].values\n",
    "# X_train, X_test, y_train, y_test = train_test_split(X_cat, y_cat, test_size = 0.3)\n",
    "# from sklearn.neighbors import KNeighborsClassifier as KNN\n",
    "# classifier = KNN()\n",
    "# classifier = classifier.fit(X_train, y_train)\n",
    "# print(classifier.score(X_train, y_train))\n",
    "# classifier.score(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 159,
   "metadata": {},
   "outputs": [],
   "source": [
    "# from sklearn.model_selection import cross_val_score\n",
    "# scores = cross_val_score(classifier, X_cat, \n",
    "#                           y_cat, scoring='accuracy', cv=10)\n",
    "# print(scores)\n",
    "# np.mean(scores)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 160,
   "metadata": {},
   "outputs": [],
   "source": [
    "# X = sc.fit_transform(X)\n",
    "X_train, X_test, y_train, y_test = train_test_split(X, y, test_size = 0.3)\n",
    "# X_train = sc.fit_transform(X_train)\n",
    "# X_test = sc.fit_transform(X_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 161,
   "metadata": {},
   "outputs": [],
   "source": [
    "# from sklearn.tree import DecisionTreeClassifier\n",
    "# classifier = DecisionTreeClassifier(max_depth=2)\n",
    "# classifier = classifier.fit(X_train[:,:3], y_train[:,-1])\n",
    "# y_pred = classifier.predict(X_test[:,:3])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 162,
   "metadata": {},
   "outputs": [],
   "source": [
    "# from sklearn.neighbors import KNeighborsClassifier as KNN\n",
    "# classifier = KNN()\n",
    "# classifier = classifier.fit(X_train[:,:3], y_train[:,-1])\n",
    "# y_pred = classifier.predict(X_test[:,:3])\n",
    "# # y_test[:,-1] = y_pred\n",
    "# # np.array(y_test[:,-1])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 163,
   "metadata": {},
   "outputs": [],
   "source": [
    "# plt.scatter(y_pred, np.array(y_test[:,-2]), alpha=0.1, c='red')\n",
    "# plt.scatter(np.array(y_test[:,-1]), np.array(y_test[:,-2]), alpha=0.1, c='blue')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 164,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9913737612140897"
      ]
     },
     "execution_count": 164,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.ensemble import RandomForestRegressor as RFR\n",
    "rf = RFR(n_estimators=500, min_samples_leaf=1)\n",
    "rf = rf.fit(X_train, y_train)\n",
    "rf.score(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 165,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.9682929160159275"
      ]
     },
     "execution_count": 165,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "rf.score(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# rf = RFR(n_estimators=500, min_samples_leaf=10)\n",
    "# rf = rf.fit(X_train, y_train)\n",
    "# rf.score(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# rf.score(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 135,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[ 0.38088039  0.9653962   0.48815529  0.90463087 -0.06028231  0.92541169\n",
      "  0.96011026  0.94727859  0.61386696  0.93981773  0.61448689  0.8083165\n",
      "  0.28777525  0.57201167  0.95857971]\n",
      "0.6870957131372262\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0.3001466792379951"
      ]
     },
     "execution_count": 135,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.model_selection import cross_val_score\n",
    "scores = cross_val_score(rf, X, \n",
    "                          y, scoring='r2', cv=15)\n",
    "print(scores)\n",
    "print(np.mean(scores))\n",
    "np.std(scores)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": [
    "# array = data[var10].isna()\n",
    "# array = array.values\n",
    "# for i in range(len(array)):\n",
    "#     if array[i]:\n",
    "#         print(i)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 136,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.99589431159824"
      ]
     },
     "execution_count": 136,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "knn.fit(X_train, y_train)\n",
    "knn.score(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 137,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.514456027106972"
      ]
     },
     "execution_count": 137,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "knn.score(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 138,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[-0.56118848 -0.47530227 -0.41319604 -0.49892246 -0.48659237]\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "-0.4870403261569841"
      ]
     },
     "execution_count": 138,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.model_selection import cross_val_score\n",
    "scores = -cross_val_score(knn, X, \n",
    "                          y, scoring='r2', cv=5)\n",
    "print(scores)\n",
    "np.mean(scores)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 139,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.6012277285297052"
      ]
     },
     "execution_count": 139,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "from sklearn.tree import DecisionTreeRegressor as DTR\n",
    "dt = DTR(min_samples_leaf=10)\n",
    "dt = dt.fit(X_train, y_train)\n",
    "dt.score(X_train, y_train)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 140,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "0.678214656453773"
      ]
     },
     "execution_count": 140,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "dt.score(X_test, y_test)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 141,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[-0.84801534 -0.59578702 -0.77606602 -0.63191474 -0.54349945]\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "-0.6790565153293878"
      ]
     },
     "execution_count": 141,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "scores = -cross_val_score(dt, X, \n",
    "                          y, scoring='r2', cv=5)\n",
    "print(scores)\n",
    "np.mean(scores)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  },
  {
   "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": "3.8.2"
  }
 },
 "nbformat": 4,
 "nbformat_minor": 4
}