House Prices, Advanced Regression Techniques -Part_2-.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 83,
   "metadata": {},
   "outputs": [],
   "source": [
    "from plot_lib import plot_data, plot_model, set_default"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 84,
   "metadata": {},
   "outputs": [],
   "source": [
    "set_default()"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Feature choosing"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 51,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "from sklearn.linear_model import SGDRegressor\n",
    "from sklearn import svm\n",
    "from xgboost import XGBRegressor,plot_importance\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import f1_score,log_loss\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 52,
   "metadata": {},
   "outputs": [],
   "source": [
    "train = pd.read_csv('train_cleaning.csv')\n",
    "label = pd.read_csv('label.csv')\n",
    "test = pd.read_csv('test_cleaning.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 53,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_str = train.select_dtypes(include=['object'])\n",
    "train_int = train.select_dtypes(exclude=['object'])\n",
    "test_str  = test.select_dtypes (include =['object'])\n",
    "test_int  = test.select_dtypes (exclude =['object'])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 54,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn import preprocessing\n",
    "le = preprocessing.LabelEncoder()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 55,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/yasser/anaconda3/envs/AMMI/lib/python3.6/site-packages/pandas/core/indexing.py:630: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  self.obj[item_labels[indexer[info_axis]]] = value\n",
      "/home/yasser/anaconda3/envs/AMMI/lib/python3.6/site-packages/pandas/core/indexing.py:543: SettingWithCopyWarning: \n",
      "A value is trying to be set on a copy of a slice from a DataFrame.\n",
      "Try using .loc[row_indexer,col_indexer] = value instead\n",
      "\n",
      "See the caveats in the documentation: http://pandas.pydata.org/pandas-docs/stable/indexing.html#indexing-view-versus-copy\n",
      "  self.obj[item] = s\n"
     ]
    }
   ],
   "source": [
    "for i in range(train_str.shape[1]):\n",
    "    le.fit(train_str.iloc[:,i])\n",
    "    train_str.iloc[:,i] = le.transform(train_str.iloc[:,i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 56,
   "metadata": {},
   "outputs": [],
   "source": [
    "train = pd.concat([train_int,train_str],axis=1)\n",
    "label = pd.read_csv('label.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 57,
   "metadata": {},
   "outputs": [],
   "source": [
    "for i in range(test_str.shape[1]):\n",
    "    le.fit(test_str.iloc[:,i])\n",
    "    test_str.iloc[:,i] = le.transform(test_str.iloc[:,i])"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 58,
   "metadata": {},
   "outputs": [],
   "source": [
    "test = pd.concat([test_int,test_str],axis=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 59,
   "metadata": {},
   "outputs": [],
   "source": [
    "train.to_csv('train_set.csv',index=False)\n",
    "test.to_csv('test_set.csv',index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 60,
   "metadata": {},
   "outputs": [],
   "source": [
    "import os\n",
    "import pandas as pd\n",
    "import numpy as np\n",
    "from sklearn.linear_model import SGDRegressor\n",
    "from sklearn import svm\n",
    "from xgboost import XGBRegressor,plot_importance\n",
    "from sklearn.model_selection import train_test_split\n",
    "from sklearn.metrics import f1_score,log_loss\n",
    "import matplotlib.pyplot as plt"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 91,
   "metadata": {},
   "outputs": [],
   "source": [
    "train = pd.read_csv('train_set.csv')\n",
    "label = pd.read_csv('label.csv')\n",
    "test = pd.read_csv('test_set.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 92,
   "metadata": {},
   "outputs": [],
   "source": [
    "train = train.drop('Id', axis=1)\n",
    "test = test.drop('Id', axis=1)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 1. Using Lasso to degenerate the parameters"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 93,
   "metadata": {},
   "outputs": [
    {
     "name": "stderr",
     "output_type": "stream",
     "text": [
      "/home/yasser/anaconda3/envs/AMMI/lib/python3.6/site-packages/sklearn/utils/validation.py:752: DataConversionWarning: A column-vector y was passed when a 1d array was expected. Please change the shape of y to (n_samples, ), for example using ravel().\n",
      "  y = column_or_1d(y, warn=True)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "SGDRegressor(alpha=0.01, average=False, early_stopping=False, epsilon=0.1,\n",
       "       eta0=0.01, fit_intercept=True, l1_ratio=0.15,\n",
       "       learning_rate='invscaling', loss='squared_loss', max_iter=30,\n",
       "       n_iter=None, n_iter_no_change=5, penalty='l1', power_t=0.25,\n",
       "       random_state=1, shuffle=True, tol=None, validation_fraction=0.1,\n",
       "       verbose=0, warm_start=False)"
      ]
     },
     "execution_count": 93,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sgd_l1 = SGDRegressor(max_iter=30, penalty=\"l1\", random_state=1,alpha=1e-2)\n",
    "sgd_l1.fit(train, label)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 94,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "array([-8.93306765e+12, -3.06630509e+12,  2.02631697e+12, -2.62824477e+11,\n",
       "       -5.55504432e+12,  5.04004425e+12,  8.72154330e+09,  7.73096455e+10,\n",
       "       -1.05383464e+12, -3.52832800e+12,  1.17209080e+11, -4.11115297e+09,\n",
       "        4.51954060e+12,  8.70751097e+09, -1.68261746e+12, -6.96768878e+12,\n",
       "       -8.11121240e+10, -1.57528419e+09, -4.35488595e+12,  3.28955859e+12,\n",
       "        5.03168396e+12,  9.24833724e+11, -9.70099057e+12,  1.45102538e+13,\n",
       "       -2.92022221e+11, -5.84565626e+12, -1.52275747e+11, -1.30402753e+11,\n",
       "        2.77864679e+12,  1.91966024e+10, -3.47363128e+11, -1.56883471e+12,\n",
       "       -2.78043153e+12, -3.15673797e+12, -1.75453444e+12,  2.64514188e+12,\n",
       "       -5.52753721e+10, -3.74685442e+10, -7.19461944e+10, -1.57754404e+11,\n",
       "       -1.92926076e+11, -2.66368611e+11,  2.52256406e+11, -5.31924199e+10,\n",
       "       -1.82247272e+11, -3.73137997e+10, -4.58062624e+10, -1.11696110e+11,\n",
       "        1.17335926e+11, -4.08371635e+11, -3.63097099e+11, -1.20512817e+11,\n",
       "       -9.95183126e+10, -6.46876771e+10, -2.80217966e+11, -6.97533696e+10,\n",
       "        4.53961707e+10,  1.70583605e+11,  2.06910857e+11,  7.67289942e+10,\n",
       "       -2.95179098e+11, -3.26380770e+11,  1.02292226e+11,  2.53875985e+11,\n",
       "       -5.87263590e+10,  2.63054879e+11, -3.73818910e+11,  2.90930293e+11,\n",
       "        1.19874826e+11,  1.30251171e+11,  3.49943229e+11, -6.98773085e+09,\n",
       "       -1.21200437e+10,  1.45486269e+11, -5.68662125e+08,  4.58974263e+11,\n",
       "       -3.02244161e+11, -1.27180356e+10,  8.33297127e+09])"
      ]
     },
     "execution_count": 94,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sgd_l1.coef_"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 95,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(-9700990569838.328, 14510253752097.865)"
      ]
     },
     "execution_count": 95,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "sgd_l1.coef_.min(), sgd_l1.coef_.max()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 96,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Features reduced from         79 to         22\n"
     ]
    }
   ],
   "source": [
    "good_feats = np.abs(np.array(sgd_l1.coef_)) > 1e+12 # lower bound alpha=1e-6\n",
    "\n",
    "print(\"Features reduced from %10d to %10d\" % (train.shape[1], int(good_feats.sum())))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 97,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_l1_select = train.loc[:, good_feats]\n",
    "test_l1_select = test.loc[:, good_feats]"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 2. Using XGBooster to plot importance"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 98,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "XGBRegressor(base_score=0.5, booster='gbtree', colsample_bylevel=1,\n",
       "       colsample_bytree=1, gamma=0, learning_rate=0.01, max_delta_step=0,\n",
       "       max_depth=5, min_child_weight=1, missing=None, n_estimators=1000,\n",
       "       n_jobs=1, nthread=None, objective='reg:linear', random_state=0,\n",
       "       reg_alpha=0, reg_lambda=1, scale_pos_weight=1, seed=None,\n",
       "       silent=True, subsample=1)"
      ]
     },
     "execution_count": 98,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = XGBRegressor(max_depth=5,n_estimators=1000,learning_rate=1e-2)\n",
    "model.fit(train,label)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 99,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0.07999649 0.03131442 0.0011403  0.01061357 0.05227841 0.00697338\n",
      " 0.00978027 0.00320161 0.03381431 0.01157844 0.0099557  0.00605237\n",
      " 0.05091882 0.00903469 0.02947239 0.07411955 0.00179817 0.00179817\n",
      " 0.06561115 0.02521819 0.00122802 0.00214903 0.01749923 0.00078944\n",
      " 0.02381475 0.02399018 0.03622648 0.02846366 0.0053945  0.01293803\n",
      " 0.0053945  0.0405684  0.02070085 0.02534977 0.02403403 0.01083286\n",
      " 0.00074558 0.         0.00175431 0.01460462 0.00631551 0.00135959\n",
      " 0.00758739 0.00364019 0.01197316 0.00214903 0.000307   0.00166659\n",
      " 0.00385948 0.01377133 0.00451735 0.00346476 0.00548222 0.0039472\n",
      " 0.010482   0.00017543 0.00236832 0.0022806  0.00451735 0.00017543\n",
      " 0.00333319 0.00412263 0.0073681  0.00390334 0.00109644 0.00640323\n",
      " 0.00416648 0.01381518 0.00495592 0.         0.02482347 0.00135959\n",
      " 0.         0.00048243 0.00346476 0.01653436 0.00298233 0.\n",
      " 0.        ]\n"
     ]
    }
   ],
   "source": [
    "print(model.feature_importances_)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 100,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "image/png": "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\n",
      "text/plain": [
       "<Figure size 720x720 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "plot_importance(model,max_num_features=25)\n",
    "plt.show()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 101,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "Features reduced from         79 to         29\n"
     ]
    }
   ],
   "source": [
    "good_feats_xgb = np.abs(np.array(model.feature_importances_)) > 1e-2\n",
    "print(\"Features reduced from %10d to %10d\" % (train.shape[1], int(good_feats_xgb.sum())))"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 102,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_xgb_select = train.loc[:, good_feats_xgb]\n",
    "test_xgb_select = test.loc[:, good_feats_xgb]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 112,
   "metadata": {},
   "outputs": [],
   "source": [
    "train_xgb_select.to_csv('train_29feautre.csv',index=False)\n",
    "test_xgb_select.to_csv('test_29feautre.csv',index=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "# Model training"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 1.lasso selection"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 103,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_train,x_val,y_train,y_val = train_test_split(train_l1_select,label,test_size = 0.3,random_state = 2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 104,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0]\tvalidation_0-rmse:198746\n",
      "Will train until validation_0-rmse hasn't improved in 200 rounds.\n",
      "[1]\tvalidation_0-rmse:196852\n",
      "[2]\tvalidation_0-rmse:194976\n",
      "[3]\tvalidation_0-rmse:193145\n",
      "[4]\tvalidation_0-rmse:191311\n",
      "[5]\tvalidation_0-rmse:189520\n",
      "[6]\tvalidation_0-rmse:187724\n",
      "[7]\tvalidation_0-rmse:185940\n",
      "[8]\tvalidation_0-rmse:184205\n",
      "[9]\tvalidation_0-rmse:182459\n",
      "[10]\tvalidation_0-rmse:180755\n",
      "[11]\tvalidation_0-rmse:179041\n",
      "[12]\tvalidation_0-rmse:177366\n",
      "[13]\tvalidation_0-rmse:175681\n",
      "[14]\tvalidation_0-rmse:174046\n",
      "[15]\tvalidation_0-rmse:172436\n",
      "[16]\tvalidation_0-rmse:170843\n",
      "[17]\tvalidation_0-rmse:169221\n",
      "[18]\tvalidation_0-rmse:167659\n",
      "[19]\tvalidation_0-rmse:166114\n",
      "[20]\tvalidation_0-rmse:164547\n",
      "[21]\tvalidation_0-rmse:163014\n",
      "[22]\tvalidation_0-rmse:161509\n",
      "[23]\tvalidation_0-rmse:160008\n",
      "[24]\tvalidation_0-rmse:158546\n",
      "[25]\tvalidation_0-rmse:157081\n",
      "[26]\tvalidation_0-rmse:155631\n",
      "[27]\tvalidation_0-rmse:154200\n",
      "[28]\tvalidation_0-rmse:152798\n",
      "[29]\tvalidation_0-rmse:151391\n",
      "[30]\tvalidation_0-rmse:150005\n",
      "[31]\tvalidation_0-rmse:148626\n",
      "[32]\tvalidation_0-rmse:147277\n",
      "[33]\tvalidation_0-rmse:145928\n",
      "[34]\tvalidation_0-rmse:144602\n",
      "[35]\tvalidation_0-rmse:143283\n",
      "[36]\tvalidation_0-rmse:141992\n",
      "[37]\tvalidation_0-rmse:140704\n",
      "[38]\tvalidation_0-rmse:139402\n",
      "[39]\tvalidation_0-rmse:138142\n",
      "[40]\tvalidation_0-rmse:136897\n",
      "[41]\tvalidation_0-rmse:135664\n",
      "[42]\tvalidation_0-rmse:134458\n",
      "[43]\tvalidation_0-rmse:133256\n",
      "[44]\tvalidation_0-rmse:132084\n",
      "[45]\tvalidation_0-rmse:130907\n",
      "[46]\tvalidation_0-rmse:129757\n",
      "[47]\tvalidation_0-rmse:128606\n",
      "[48]\tvalidation_0-rmse:127486\n",
      "[49]\tvalidation_0-rmse:126375\n",
      "[50]\tvalidation_0-rmse:125263\n",
      "[51]\tvalidation_0-rmse:124158\n",
      "[52]\tvalidation_0-rmse:123080\n",
      "[53]\tvalidation_0-rmse:122019\n",
      "[54]\tvalidation_0-rmse:120956\n",
      "[55]\tvalidation_0-rmse:119901\n",
      "[56]\tvalidation_0-rmse:118873\n",
      "[57]\tvalidation_0-rmse:117852\n",
      "[58]\tvalidation_0-rmse:116836\n",
      "[59]\tvalidation_0-rmse:115832\n",
      "[60]\tvalidation_0-rmse:114844\n",
      "[61]\tvalidation_0-rmse:113877\n",
      "[62]\tvalidation_0-rmse:112909\n",
      "[63]\tvalidation_0-rmse:111948\n",
      "[64]\tvalidation_0-rmse:110996\n",
      "[65]\tvalidation_0-rmse:110061\n",
      "[66]\tvalidation_0-rmse:109143\n",
      "[67]\tvalidation_0-rmse:108191\n",
      "[68]\tvalidation_0-rmse:107294\n",
      "[69]\tvalidation_0-rmse:106398\n",
      "[70]\tvalidation_0-rmse:105519\n",
      "[71]\tvalidation_0-rmse:104653\n",
      "[72]\tvalidation_0-rmse:103769\n",
      "[73]\tvalidation_0-rmse:102916\n",
      "[74]\tvalidation_0-rmse:102077\n",
      "[75]\tvalidation_0-rmse:101243\n",
      "[76]\tvalidation_0-rmse:100412\n",
      "[77]\tvalidation_0-rmse:99601\n",
      "[78]\tvalidation_0-rmse:98788.3\n",
      "[79]\tvalidation_0-rmse:98002.3\n",
      "[80]\tvalidation_0-rmse:97229.9\n",
      "[81]\tvalidation_0-rmse:96469.2\n",
      "[82]\tvalidation_0-rmse:95708.2\n",
      "[83]\tvalidation_0-rmse:94937.6\n",
      "[84]\tvalidation_0-rmse:94176.6\n",
      "[85]\tvalidation_0-rmse:93395.5\n",
      "[86]\tvalidation_0-rmse:92651.7\n",
      "[87]\tvalidation_0-rmse:91924.3\n",
      "[88]\tvalidation_0-rmse:91184.7\n",
      "[89]\tvalidation_0-rmse:90477.1\n",
      "[90]\tvalidation_0-rmse:89741.6\n",
      "[91]\tvalidation_0-rmse:89049\n",
      "[92]\tvalidation_0-rmse:88356.8\n",
      "[93]\tvalidation_0-rmse:87645.1\n",
      "[94]\tvalidation_0-rmse:86969.9\n",
      "[95]\tvalidation_0-rmse:86304.3\n",
      "[96]\tvalidation_0-rmse:85652.7\n",
      "[97]\tvalidation_0-rmse:85011\n",
      "[98]\tvalidation_0-rmse:84337.1\n",
      "[99]\tvalidation_0-rmse:83702.4\n",
      "[100]\tvalidation_0-rmse:83075.4\n",
      "[101]\tvalidation_0-rmse:82457\n",
      "[102]\tvalidation_0-rmse:81851.3\n",
      "[103]\tvalidation_0-rmse:81252.4\n",
      "[104]\tvalidation_0-rmse:80652.5\n",
      "[105]\tvalidation_0-rmse:80031\n",
      "[106]\tvalidation_0-rmse:79448.6\n",
      "[107]\tvalidation_0-rmse:78876.5\n",
      "[108]\tvalidation_0-rmse:78314.2\n",
      "[109]\tvalidation_0-rmse:77760\n",
      "[110]\tvalidation_0-rmse:77212.7\n",
      "[111]\tvalidation_0-rmse:76657.8\n",
      "[112]\tvalidation_0-rmse:76084.2\n",
      "[113]\tvalidation_0-rmse:75519.5\n",
      "[114]\tvalidation_0-rmse:74991.5\n",
      "[115]\tvalidation_0-rmse:74481\n",
      "[116]\tvalidation_0-rmse:73934.4\n",
      "[117]\tvalidation_0-rmse:73389.3\n",
      "[118]\tvalidation_0-rmse:72887.8\n",
      "[119]\tvalidation_0-rmse:72397.1\n",
      "[120]\tvalidation_0-rmse:71910.1\n",
      "[121]\tvalidation_0-rmse:71430\n",
      "[122]\tvalidation_0-rmse:70926.4\n",
      "[123]\tvalidation_0-rmse:70459.6\n",
      "[124]\tvalidation_0-rmse:69996.8\n",
      "[125]\tvalidation_0-rmse:69510\n",
      "[126]\tvalidation_0-rmse:69063.2\n",
      "[127]\tvalidation_0-rmse:68618.9\n",
      "[128]\tvalidation_0-rmse:68190.1\n",
      "[129]\tvalidation_0-rmse:67762\n",
      "[130]\tvalidation_0-rmse:67337.5\n",
      "[131]\tvalidation_0-rmse:66912.6\n",
      "[132]\tvalidation_0-rmse:66491.1\n",
      "[133]\tvalidation_0-rmse:66083.9\n",
      "[134]\tvalidation_0-rmse:65683.6\n",
      "[135]\tvalidation_0-rmse:65280.4\n",
      "[136]\tvalidation_0-rmse:64884.6\n",
      "[137]\tvalidation_0-rmse:64504.9\n",
      "[138]\tvalidation_0-rmse:64126.8\n",
      "[139]\tvalidation_0-rmse:63745.2\n",
      "[140]\tvalidation_0-rmse:63378.7\n",
      "[141]\tvalidation_0-rmse:63019.4\n",
      "[142]\tvalidation_0-rmse:62665.2\n",
      "[143]\tvalidation_0-rmse:62310.8\n",
      "[144]\tvalidation_0-rmse:61961.1\n",
      "[145]\tvalidation_0-rmse:61625\n",
      "[146]\tvalidation_0-rmse:61290.6\n",
      "[147]\tvalidation_0-rmse:60948.7\n",
      "[148]\tvalidation_0-rmse:60616.5\n",
      "[149]\tvalidation_0-rmse:60293.1\n",
      "[150]\tvalidation_0-rmse:59968.4\n",
      "[151]\tvalidation_0-rmse:59664.3\n",
      "[152]\tvalidation_0-rmse:59350.6\n",
      "[153]\tvalidation_0-rmse:59023.2\n",
      "[154]\tvalidation_0-rmse:58719.3\n",
      "[155]\tvalidation_0-rmse:58417.1\n",
      "[156]\tvalidation_0-rmse:58105.8\n",
      "[157]\tvalidation_0-rmse:57829.2\n",
      "[158]\tvalidation_0-rmse:57538.9\n",
      "[159]\tvalidation_0-rmse:57253.7\n",
      "[160]\tvalidation_0-rmse:56972.7\n",
      "[161]\tvalidation_0-rmse:56682.4\n",
      "[162]\tvalidation_0-rmse:56427.6\n",
      "[163]\tvalidation_0-rmse:56157\n",
      "[164]\tvalidation_0-rmse:55881.5\n",
      "[165]\tvalidation_0-rmse:55600.9\n",
      "[166]\tvalidation_0-rmse:55359.5\n",
      "[167]\tvalidation_0-rmse:55088.2\n",
      "[168]\tvalidation_0-rmse:54829.9\n",
      "[169]\tvalidation_0-rmse:54604.9\n",
      "[170]\tvalidation_0-rmse:54345.6\n",
      "[171]\tvalidation_0-rmse:54124.2\n",
      "[172]\tvalidation_0-rmse:53883.8\n",
      "[173]\tvalidation_0-rmse:53637.2\n",
      "[174]\tvalidation_0-rmse:53430.1\n",
      "[175]\tvalidation_0-rmse:53191.8\n",
      "[176]\tvalidation_0-rmse:52956.9\n",
      "[177]\tvalidation_0-rmse:52725\n",
      "[178]\tvalidation_0-rmse:52530.4\n",
      "[179]\tvalidation_0-rmse:52303\n",
      "[180]\tvalidation_0-rmse:52091.2\n",
      "[181]\tvalidation_0-rmse:51906.8\n",
      "[182]\tvalidation_0-rmse:51708.8\n",
      "[183]\tvalidation_0-rmse:51497.5\n",
      "[184]\tvalidation_0-rmse:51324.6\n",
      "[185]\tvalidation_0-rmse:51119.2\n",
      "[186]\tvalidation_0-rmse:50919.1\n",
      "[187]\tvalidation_0-rmse:50737.6\n",
      "[188]\tvalidation_0-rmse:50545.1\n",
      "[189]\tvalidation_0-rmse:50377.9\n",
      "[190]\tvalidation_0-rmse:50190.6\n",
      "[191]\tvalidation_0-rmse:50020.6\n",
      "[192]\tvalidation_0-rmse:49841.6\n",
      "[193]\tvalidation_0-rmse:49691\n",
      "[194]\tvalidation_0-rmse:49517.3\n",
      "[195]\tvalidation_0-rmse:49346\n",
      "[196]\tvalidation_0-rmse:49190.9\n",
      "[197]\tvalidation_0-rmse:49046.6\n",
      "[198]\tvalidation_0-rmse:48884.6\n",
      "[199]\tvalidation_0-rmse:48724.2\n",
      "[200]\tvalidation_0-rmse:48588.1\n",
      "[201]\tvalidation_0-rmse:48434.8\n",
      "[202]\tvalidation_0-rmse:48301\n",
      "[203]\tvalidation_0-rmse:48154.2\n",
      "[204]\tvalidation_0-rmse:48007.6\n",
      "[205]\tvalidation_0-rmse:47881.9\n",
      "[206]\tvalidation_0-rmse:47740.1\n",
      "[207]\tvalidation_0-rmse:47603.2\n",
      "[208]\tvalidation_0-rmse:47487.8\n",
      "[209]\tvalidation_0-rmse:47356.8\n",
      "[210]\tvalidation_0-rmse:47231.4\n",
      "[211]\tvalidation_0-rmse:47114\n",
      "[212]\tvalidation_0-rmse:46991.6\n",
      "[213]\tvalidation_0-rmse:46880.3\n",
      "[214]\tvalidation_0-rmse:46770\n",
      "[215]\tvalidation_0-rmse:46654.5\n",
      "[216]\tvalidation_0-rmse:46541.1\n",
      "[217]\tvalidation_0-rmse:46435.7\n",
      "[218]\tvalidation_0-rmse:46331.4\n",
      "[219]\tvalidation_0-rmse:46222.3\n",
      "[220]\tvalidation_0-rmse:46123.5\n",
      "[221]\tvalidation_0-rmse:46023.6\n",
      "[222]\tvalidation_0-rmse:45937.3\n",
      "[223]\tvalidation_0-rmse:45844.1\n",
      "[224]\tvalidation_0-rmse:45762.9\n",
      "[225]\tvalidation_0-rmse:45681.7\n",
      "[226]\tvalidation_0-rmse:45592.9\n",
      "[227]\tvalidation_0-rmse:45512.9\n",
      "[228]\tvalidation_0-rmse:45417.3\n",
      "[229]\tvalidation_0-rmse:45344.6\n",
      "[230]\tvalidation_0-rmse:45265.2\n",
      "[231]\tvalidation_0-rmse:45186.2\n",
      "[232]\tvalidation_0-rmse:45104.6\n",
      "[233]\tvalidation_0-rmse:45032.7\n",
      "[234]\tvalidation_0-rmse:44958.7\n",
      "[235]\tvalidation_0-rmse:44886.3\n",
      "[236]\tvalidation_0-rmse:44824.1\n",
      "[237]\tvalidation_0-rmse:44745.5\n",
      "[238]\tvalidation_0-rmse:44681.8\n",
      "[239]\tvalidation_0-rmse:44637\n",
      "[240]\tvalidation_0-rmse:44580.4\n",
      "[241]\tvalidation_0-rmse:44523\n",
      "[242]\tvalidation_0-rmse:44484.4\n",
      "[243]\tvalidation_0-rmse:44432.3\n",
      "[244]\tvalidation_0-rmse:44390\n",
      "[245]\tvalidation_0-rmse:44334\n",
      "[246]\tvalidation_0-rmse:44293.5\n",
      "[247]\tvalidation_0-rmse:44239.5\n",
      "[248]\tvalidation_0-rmse:44191.9\n",
      "[249]\tvalidation_0-rmse:44162.6\n",
      "[250]\tvalidation_0-rmse:44116.6\n",
      "[251]\tvalidation_0-rmse:44083.2\n",
      "[252]\tvalidation_0-rmse:44037.6\n",
      "[253]\tvalidation_0-rmse:43995.7\n",
      "[254]\tvalidation_0-rmse:43967\n",
      "[255]\tvalidation_0-rmse:43925.4\n",
      "[256]\tvalidation_0-rmse:43886.4\n",
      "[257]\tvalidation_0-rmse:43851\n",
      "[258]\tvalidation_0-rmse:43814.7\n",
      "[259]\tvalidation_0-rmse:43778.5\n",
      "[260]\tvalidation_0-rmse:43746.8\n",
      "[261]\tvalidation_0-rmse:43713.2\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[262]\tvalidation_0-rmse:43693.1\n",
      "[263]\tvalidation_0-rmse:43660.6\n",
      "[264]\tvalidation_0-rmse:43629.1\n",
      "[265]\tvalidation_0-rmse:43599.3\n",
      "[266]\tvalidation_0-rmse:43570\n",
      "[267]\tvalidation_0-rmse:43543.7\n",
      "[268]\tvalidation_0-rmse:43518.3\n",
      "[269]\tvalidation_0-rmse:43505.2\n",
      "[270]\tvalidation_0-rmse:43482.3\n",
      "[271]\tvalidation_0-rmse:43458.5\n",
      "[272]\tvalidation_0-rmse:43437.6\n",
      "[273]\tvalidation_0-rmse:43412.1\n",
      "[274]\tvalidation_0-rmse:43392.9\n",
      "[275]\tvalidation_0-rmse:43371.6\n",
      "[276]\tvalidation_0-rmse:43365.2\n",
      "[277]\tvalidation_0-rmse:43345.3\n",
      "[278]\tvalidation_0-rmse:43326.6\n",
      "[279]\tvalidation_0-rmse:43308.1\n",
      "[280]\tvalidation_0-rmse:43289.6\n",
      "[281]\tvalidation_0-rmse:43286.8\n",
      "[282]\tvalidation_0-rmse:43268.5\n",
      "[283]\tvalidation_0-rmse:43248.2\n",
      "[284]\tvalidation_0-rmse:43232.9\n",
      "[285]\tvalidation_0-rmse:43220.9\n",
      "[286]\tvalidation_0-rmse:43196.5\n",
      "[287]\tvalidation_0-rmse:43178.5\n",
      "[288]\tvalidation_0-rmse:43168.7\n",
      "[289]\tvalidation_0-rmse:43146.2\n",
      "[290]\tvalidation_0-rmse:43123.2\n",
      "[291]\tvalidation_0-rmse:43111\n",
      "[292]\tvalidation_0-rmse:43090.8\n",
      "[293]\tvalidation_0-rmse:43085.5\n",
      "[294]\tvalidation_0-rmse:43067.6\n",
      "[295]\tvalidation_0-rmse:43051.9\n",
      "[296]\tvalidation_0-rmse:43036.7\n",
      "[297]\tvalidation_0-rmse:43032.9\n",
      "[298]\tvalidation_0-rmse:43015.7\n",
      "[299]\tvalidation_0-rmse:43003.3\n",
      "[300]\tvalidation_0-rmse:43000.4\n",
      "[301]\tvalidation_0-rmse:42986.8\n",
      "[302]\tvalidation_0-rmse:42973.7\n",
      "[303]\tvalidation_0-rmse:42963.6\n",
      "[304]\tvalidation_0-rmse:42949.6\n",
      "[305]\tvalidation_0-rmse:42945.8\n",
      "[306]\tvalidation_0-rmse:42936.4\n",
      "[307]\tvalidation_0-rmse:42932.4\n",
      "[308]\tvalidation_0-rmse:42924.3\n",
      "[309]\tvalidation_0-rmse:42914\n",
      "[310]\tvalidation_0-rmse:42911.1\n",
      "[311]\tvalidation_0-rmse:42904.7\n",
      "[312]\tvalidation_0-rmse:42897.5\n",
      "[313]\tvalidation_0-rmse:42891.7\n",
      "[314]\tvalidation_0-rmse:42899\n",
      "[315]\tvalidation_0-rmse:42893.4\n",
      "[316]\tvalidation_0-rmse:42904.5\n",
      "[317]\tvalidation_0-rmse:42904.3\n",
      "[318]\tvalidation_0-rmse:42898.7\n",
      "[319]\tvalidation_0-rmse:42887.8\n",
      "[320]\tvalidation_0-rmse:42877.3\n",
      "[321]\tvalidation_0-rmse:42872.9\n",
      "[322]\tvalidation_0-rmse:42875.3\n",
      "[323]\tvalidation_0-rmse:42864.2\n",
      "[324]\tvalidation_0-rmse:42855\n",
      "[325]\tvalidation_0-rmse:42847\n",
      "[326]\tvalidation_0-rmse:42838.9\n",
      "[327]\tvalidation_0-rmse:42840.8\n",
      "[328]\tvalidation_0-rmse:42831.6\n",
      "[329]\tvalidation_0-rmse:42833.3\n",
      "[330]\tvalidation_0-rmse:42835.4\n",
      "[331]\tvalidation_0-rmse:42826.3\n",
      "[332]\tvalidation_0-rmse:42829.6\n",
      "[333]\tvalidation_0-rmse:42820.2\n",
      "[334]\tvalidation_0-rmse:42823.1\n",
      "[335]\tvalidation_0-rmse:42813\n",
      "[336]\tvalidation_0-rmse:42816.6\n",
      "[337]\tvalidation_0-rmse:42807.1\n",
      "[338]\tvalidation_0-rmse:42811.3\n",
      "[339]\tvalidation_0-rmse:42802\n",
      "[340]\tvalidation_0-rmse:42806.3\n",
      "[341]\tvalidation_0-rmse:42799.2\n",
      "[342]\tvalidation_0-rmse:42804\n",
      "[343]\tvalidation_0-rmse:42795.8\n",
      "[344]\tvalidation_0-rmse:42796.9\n",
      "[345]\tvalidation_0-rmse:42790.6\n",
      "[346]\tvalidation_0-rmse:42790.7\n",
      "[347]\tvalidation_0-rmse:42795.6\n",
      "[348]\tvalidation_0-rmse:42792.1\n",
      "[349]\tvalidation_0-rmse:42798.4\n",
      "[350]\tvalidation_0-rmse:42813.6\n",
      "[351]\tvalidation_0-rmse:42827.3\n",
      "[352]\tvalidation_0-rmse:42842.9\n",
      "[353]\tvalidation_0-rmse:42859.2\n",
      "[354]\tvalidation_0-rmse:42872.1\n",
      "[355]\tvalidation_0-rmse:42887.8\n",
      "[356]\tvalidation_0-rmse:42904\n",
      "[357]\tvalidation_0-rmse:42919.9\n",
      "[358]\tvalidation_0-rmse:42936\n",
      "[359]\tvalidation_0-rmse:42940.3\n",
      "[360]\tvalidation_0-rmse:42954.1\n",
      "[361]\tvalidation_0-rmse:42970.7\n",
      "[362]\tvalidation_0-rmse:42986.6\n",
      "[363]\tvalidation_0-rmse:43002.8\n",
      "[364]\tvalidation_0-rmse:43007.8\n",
      "[365]\tvalidation_0-rmse:43023.4\n",
      "[366]\tvalidation_0-rmse:43039.2\n",
      "[367]\tvalidation_0-rmse:43058\n",
      "[368]\tvalidation_0-rmse:43075.5\n",
      "[369]\tvalidation_0-rmse:43082.9\n",
      "[370]\tvalidation_0-rmse:43098.6\n",
      "[371]\tvalidation_0-rmse:43116.5\n",
      "[372]\tvalidation_0-rmse:43123.8\n",
      "[373]\tvalidation_0-rmse:43140.7\n",
      "[374]\tvalidation_0-rmse:43158.8\n",
      "[375]\tvalidation_0-rmse:43155.7\n",
      "[376]\tvalidation_0-rmse:43160.4\n",
      "[377]\tvalidation_0-rmse:43157.5\n",
      "[378]\tvalidation_0-rmse:43162.5\n",
      "[379]\tvalidation_0-rmse:43181.6\n",
      "[380]\tvalidation_0-rmse:43181\n",
      "[381]\tvalidation_0-rmse:43188.9\n",
      "[382]\tvalidation_0-rmse:43206\n",
      "[383]\tvalidation_0-rmse:43213.1\n",
      "[384]\tvalidation_0-rmse:43211.1\n",
      "[385]\tvalidation_0-rmse:43216.7\n",
      "[386]\tvalidation_0-rmse:43215.1\n",
      "[387]\tvalidation_0-rmse:43227.9\n",
      "[388]\tvalidation_0-rmse:43235\n",
      "[389]\tvalidation_0-rmse:43253.1\n",
      "[390]\tvalidation_0-rmse:43253\n",
      "[391]\tvalidation_0-rmse:43265.9\n",
      "[392]\tvalidation_0-rmse:43282.6\n",
      "[393]\tvalidation_0-rmse:43281.2\n",
      "[394]\tvalidation_0-rmse:43283.1\n",
      "[395]\tvalidation_0-rmse:43280.3\n",
      "[396]\tvalidation_0-rmse:43278.7\n",
      "[397]\tvalidation_0-rmse:43289.9\n",
      "[398]\tvalidation_0-rmse:43301.1\n",
      "[399]\tvalidation_0-rmse:43318.8\n",
      "[400]\tvalidation_0-rmse:43332.3\n",
      "[401]\tvalidation_0-rmse:43331.5\n",
      "[402]\tvalidation_0-rmse:43336.3\n",
      "[403]\tvalidation_0-rmse:43340.3\n",
      "[404]\tvalidation_0-rmse:43351.7\n",
      "[405]\tvalidation_0-rmse:43355.8\n",
      "[406]\tvalidation_0-rmse:43355.2\n",
      "[407]\tvalidation_0-rmse:43366.8\n",
      "[408]\tvalidation_0-rmse:43372.7\n",
      "[409]\tvalidation_0-rmse:43380.5\n",
      "[410]\tvalidation_0-rmse:43390.9\n",
      "[411]\tvalidation_0-rmse:43397\n",
      "[412]\tvalidation_0-rmse:43410.1\n",
      "[413]\tvalidation_0-rmse:43409.4\n",
      "[414]\tvalidation_0-rmse:43412.5\n",
      "[415]\tvalidation_0-rmse:43415.6\n",
      "[416]\tvalidation_0-rmse:43417.6\n",
      "[417]\tvalidation_0-rmse:43425.6\n",
      "[418]\tvalidation_0-rmse:43438.4\n",
      "[419]\tvalidation_0-rmse:43441.7\n",
      "[420]\tvalidation_0-rmse:43448.6\n",
      "[421]\tvalidation_0-rmse:43448.5\n",
      "[422]\tvalidation_0-rmse:43448.8\n",
      "[423]\tvalidation_0-rmse:43452.4\n",
      "[424]\tvalidation_0-rmse:43460.7\n",
      "[425]\tvalidation_0-rmse:43458.8\n",
      "[426]\tvalidation_0-rmse:43457.3\n",
      "[427]\tvalidation_0-rmse:43461\n",
      "[428]\tvalidation_0-rmse:43470.8\n",
      "[429]\tvalidation_0-rmse:43475.1\n",
      "[430]\tvalidation_0-rmse:43475.2\n",
      "[431]\tvalidation_0-rmse:43479.5\n",
      "[432]\tvalidation_0-rmse:43483\n",
      "[433]\tvalidation_0-rmse:43486.4\n",
      "[434]\tvalidation_0-rmse:43485.5\n",
      "[435]\tvalidation_0-rmse:43484.2\n",
      "[436]\tvalidation_0-rmse:43486.7\n",
      "[437]\tvalidation_0-rmse:43491\n",
      "[438]\tvalidation_0-rmse:43491.1\n",
      "[439]\tvalidation_0-rmse:43494.8\n",
      "[440]\tvalidation_0-rmse:43500.4\n",
      "[441]\tvalidation_0-rmse:43500.7\n",
      "[442]\tvalidation_0-rmse:43504.6\n",
      "[443]\tvalidation_0-rmse:43505\n",
      "[444]\tvalidation_0-rmse:43506.9\n",
      "[445]\tvalidation_0-rmse:43512.5\n",
      "[446]\tvalidation_0-rmse:43515.5\n",
      "[447]\tvalidation_0-rmse:43517.2\n",
      "[448]\tvalidation_0-rmse:43520.6\n",
      "[449]\tvalidation_0-rmse:43522.9\n",
      "[450]\tvalidation_0-rmse:43523.6\n",
      "[451]\tvalidation_0-rmse:43529.9\n",
      "[452]\tvalidation_0-rmse:43535.2\n",
      "[453]\tvalidation_0-rmse:43544\n",
      "[454]\tvalidation_0-rmse:43546.7\n",
      "[455]\tvalidation_0-rmse:43547.4\n",
      "[456]\tvalidation_0-rmse:43550.6\n",
      "[457]\tvalidation_0-rmse:43553.4\n",
      "[458]\tvalidation_0-rmse:43554\n",
      "[459]\tvalidation_0-rmse:43554.9\n",
      "[460]\tvalidation_0-rmse:43560.2\n",
      "[461]\tvalidation_0-rmse:43562.9\n",
      "[462]\tvalidation_0-rmse:43572.3\n",
      "[463]\tvalidation_0-rmse:43575.9\n",
      "[464]\tvalidation_0-rmse:43576.7\n",
      "[465]\tvalidation_0-rmse:43579.5\n",
      "[466]\tvalidation_0-rmse:43588.8\n",
      "[467]\tvalidation_0-rmse:43595\n",
      "[468]\tvalidation_0-rmse:43599.8\n",
      "[469]\tvalidation_0-rmse:43601.6\n",
      "[470]\tvalidation_0-rmse:43606.8\n",
      "[471]\tvalidation_0-rmse:43612.8\n",
      "[472]\tvalidation_0-rmse:43622\n",
      "[473]\tvalidation_0-rmse:43624.1\n",
      "[474]\tvalidation_0-rmse:43633.2\n",
      "[475]\tvalidation_0-rmse:43635.1\n",
      "[476]\tvalidation_0-rmse:43641.4\n",
      "[477]\tvalidation_0-rmse:43645.4\n",
      "[478]\tvalidation_0-rmse:43649.8\n",
      "[479]\tvalidation_0-rmse:43662.2\n",
      "[480]\tvalidation_0-rmse:43664.7\n",
      "[481]\tvalidation_0-rmse:43664.7\n",
      "[482]\tvalidation_0-rmse:43666.6\n",
      "[483]\tvalidation_0-rmse:43675.7\n",
      "[484]\tvalidation_0-rmse:43679.2\n",
      "[485]\tvalidation_0-rmse:43679\n",
      "[486]\tvalidation_0-rmse:43685.6\n",
      "[487]\tvalidation_0-rmse:43697\n",
      "[488]\tvalidation_0-rmse:43697.3\n",
      "[489]\tvalidation_0-rmse:43704.5\n",
      "[490]\tvalidation_0-rmse:43711.4\n",
      "[491]\tvalidation_0-rmse:43708.7\n",
      "[492]\tvalidation_0-rmse:43714.8\n",
      "[493]\tvalidation_0-rmse:43714.9\n",
      "[494]\tvalidation_0-rmse:43713.7\n",
      "[495]\tvalidation_0-rmse:43718.9\n",
      "[496]\tvalidation_0-rmse:43716.2\n",
      "[497]\tvalidation_0-rmse:43720.8\n",
      "[498]\tvalidation_0-rmse:43719.7\n",
      "[499]\tvalidation_0-rmse:43722.1\n",
      "[500]\tvalidation_0-rmse:43731.1\n",
      "[501]\tvalidation_0-rmse:43734.4\n",
      "[502]\tvalidation_0-rmse:43738.7\n",
      "[503]\tvalidation_0-rmse:43738.9\n",
      "[504]\tvalidation_0-rmse:43739.2\n",
      "[505]\tvalidation_0-rmse:43744.3\n",
      "[506]\tvalidation_0-rmse:43746.9\n",
      "[507]\tvalidation_0-rmse:43754.2\n",
      "[508]\tvalidation_0-rmse:43755.9\n",
      "[509]\tvalidation_0-rmse:43762.1\n",
      "[510]\tvalidation_0-rmse:43769.1\n",
      "[511]\tvalidation_0-rmse:43771.5\n",
      "[512]\tvalidation_0-rmse:43775.9\n",
      "[513]\tvalidation_0-rmse:43775.3\n",
      "[514]\tvalidation_0-rmse:43775.5\n",
      "[515]\tvalidation_0-rmse:43782.6\n",
      "[516]\tvalidation_0-rmse:43782.7\n",
      "[517]\tvalidation_0-rmse:43787.4\n",
      "[518]\tvalidation_0-rmse:43785.2\n",
      "[519]\tvalidation_0-rmse:43788.8\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[520]\tvalidation_0-rmse:43791.6\n",
      "[521]\tvalidation_0-rmse:43796.5\n",
      "[522]\tvalidation_0-rmse:43799.8\n",
      "[523]\tvalidation_0-rmse:43799.9\n",
      "[524]\tvalidation_0-rmse:43801.7\n",
      "[525]\tvalidation_0-rmse:43805\n",
      "[526]\tvalidation_0-rmse:43807.3\n",
      "[527]\tvalidation_0-rmse:43809\n",
      "[528]\tvalidation_0-rmse:43812.9\n",
      "[529]\tvalidation_0-rmse:43813.2\n",
      "[530]\tvalidation_0-rmse:43816.1\n",
      "[531]\tvalidation_0-rmse:43817.6\n",
      "[532]\tvalidation_0-rmse:43820.5\n",
      "[533]\tvalidation_0-rmse:43820.4\n",
      "[534]\tvalidation_0-rmse:43824.7\n",
      "[535]\tvalidation_0-rmse:43825.8\n",
      "[536]\tvalidation_0-rmse:43827.6\n",
      "[537]\tvalidation_0-rmse:43830.8\n",
      "[538]\tvalidation_0-rmse:43833.1\n",
      "[539]\tvalidation_0-rmse:43834.7\n",
      "[540]\tvalidation_0-rmse:43837.1\n",
      "[541]\tvalidation_0-rmse:43840.2\n",
      "[542]\tvalidation_0-rmse:43842.5\n",
      "[543]\tvalidation_0-rmse:43848.8\n",
      "[544]\tvalidation_0-rmse:43852.5\n",
      "[545]\tvalidation_0-rmse:43853.7\n",
      "Stopping. Best iteration:\n",
      "[345]\tvalidation_0-rmse:42790.6\n",
      "\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "XGBRegressor(base_score=0.5, booster='gbtree', colsample_bylevel=1,\n",
       "       colsample_bytree=1, gamma=0, learning_rate=0.01, max_delta_step=0,\n",
       "       max_depth=5, min_child_weight=1, missing=None, n_estimators=1000,\n",
       "       n_jobs=1, nthread=None, objective='reg:linear', random_state=0,\n",
       "       reg_alpha=0, reg_lambda=1, scale_pos_weight=1, seed=None,\n",
       "       silent=True, subsample=1)"
      ]
     },
     "execution_count": 104,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = XGBRegressor(max_depth=5,learning_rate=1e-2,n_estimators=1000)\n",
    "model.fit(x_train,y_train,\\\n",
    "          eval_metric='rmse',\\\n",
    "          eval_set=[(x_val,y_val)],\\\n",
    "          early_stopping_rounds=200,\\\n",
    "          verbose=1)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 105,
   "metadata": {},
   "outputs": [],
   "source": [
    "pred = model.predict(test_l1_select)\n",
    "submission = pd.read_csv('sample_submission.csv')\n",
    "submission.SalePrice = pred\n",
    "submission.to_csv('submission_l1_select.csv',index=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### 2. xgb selection"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 106,
   "metadata": {},
   "outputs": [],
   "source": [
    "x_train,x_val,y_train,y_val = train_test_split(train_xgb_select,label,test_size = 0.3,random_state = 3)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 107,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[0]\tvalidation_0-rmse:191543\n",
      "Will train until validation_0-rmse hasn't improved in 200 rounds.\n",
      "[2]\tvalidation_0-rmse:187900\n",
      "[4]\tvalidation_0-rmse:184328\n",
      "[6]\tvalidation_0-rmse:180811\n",
      "[8]\tvalidation_0-rmse:177393\n",
      "[10]\tvalidation_0-rmse:174023\n",
      "[12]\tvalidation_0-rmse:170737\n",
      "[14]\tvalidation_0-rmse:167499\n",
      "[16]\tvalidation_0-rmse:164319\n",
      "[18]\tvalidation_0-rmse:161213\n",
      "[20]\tvalidation_0-rmse:158157\n",
      "[22]\tvalidation_0-rmse:155193\n",
      "[24]\tvalidation_0-rmse:152251\n",
      "[26]\tvalidation_0-rmse:149424\n",
      "[28]\tvalidation_0-rmse:146627\n",
      "[30]\tvalidation_0-rmse:143917\n",
      "[32]\tvalidation_0-rmse:141255\n",
      "[34]\tvalidation_0-rmse:138625\n",
      "[36]\tvalidation_0-rmse:136068\n",
      "[38]\tvalidation_0-rmse:133564\n",
      "[40]\tvalidation_0-rmse:131097\n",
      "[42]\tvalidation_0-rmse:128672\n",
      "[44]\tvalidation_0-rmse:126307\n",
      "[46]\tvalidation_0-rmse:123947\n",
      "[48]\tvalidation_0-rmse:121683\n",
      "[50]\tvalidation_0-rmse:119465\n",
      "[52]\tvalidation_0-rmse:117334\n",
      "[54]\tvalidation_0-rmse:115220\n",
      "[56]\tvalidation_0-rmse:113176\n",
      "[58]\tvalidation_0-rmse:111170\n",
      "[60]\tvalidation_0-rmse:109178\n",
      "[62]\tvalidation_0-rmse:107256\n",
      "[64]\tvalidation_0-rmse:105338\n",
      "[66]\tvalidation_0-rmse:103486\n",
      "[68]\tvalidation_0-rmse:101647\n",
      "[70]\tvalidation_0-rmse:99846.3\n",
      "[72]\tvalidation_0-rmse:98082.6\n",
      "[74]\tvalidation_0-rmse:96358.3\n",
      "[76]\tvalidation_0-rmse:94657.1\n",
      "[78]\tvalidation_0-rmse:93003.9\n",
      "[80]\tvalidation_0-rmse:91386.1\n",
      "[82]\tvalidation_0-rmse:89809.3\n",
      "[84]\tvalidation_0-rmse:88239.7\n",
      "[86]\tvalidation_0-rmse:86714.6\n",
      "[88]\tvalidation_0-rmse:85248.1\n",
      "[90]\tvalidation_0-rmse:83753.5\n",
      "[92]\tvalidation_0-rmse:82345.4\n",
      "[94]\tvalidation_0-rmse:80929.8\n",
      "[96]\tvalidation_0-rmse:79555.9\n",
      "[98]\tvalidation_0-rmse:78219\n",
      "[100]\tvalidation_0-rmse:76905.6\n",
      "[102]\tvalidation_0-rmse:75639.7\n",
      "[104]\tvalidation_0-rmse:74408.1\n",
      "[106]\tvalidation_0-rmse:73187\n",
      "[108]\tvalidation_0-rmse:71978.8\n",
      "[110]\tvalidation_0-rmse:70813.7\n",
      "[112]\tvalidation_0-rmse:69673\n",
      "[114]\tvalidation_0-rmse:68555\n",
      "[116]\tvalidation_0-rmse:67421.3\n",
      "[118]\tvalidation_0-rmse:66355.1\n",
      "[120]\tvalidation_0-rmse:65275.6\n",
      "[122]\tvalidation_0-rmse:64247.8\n",
      "[124]\tvalidation_0-rmse:63227.6\n",
      "[126]\tvalidation_0-rmse:62234.4\n",
      "[128]\tvalidation_0-rmse:61273.4\n",
      "[130]\tvalidation_0-rmse:60301.1\n",
      "[132]\tvalidation_0-rmse:59367.7\n",
      "[134]\tvalidation_0-rmse:58450.7\n",
      "[136]\tvalidation_0-rmse:57567\n",
      "[138]\tvalidation_0-rmse:56687.2\n",
      "[140]\tvalidation_0-rmse:55844.1\n",
      "[142]\tvalidation_0-rmse:55012.1\n",
      "[144]\tvalidation_0-rmse:54204.8\n",
      "[146]\tvalidation_0-rmse:53406.6\n",
      "[148]\tvalidation_0-rmse:52628.5\n",
      "[150]\tvalidation_0-rmse:51868.8\n",
      "[152]\tvalidation_0-rmse:51139.2\n",
      "[154]\tvalidation_0-rmse:50425.3\n",
      "[156]\tvalidation_0-rmse:49742.9\n",
      "[158]\tvalidation_0-rmse:49063.7\n",
      "[160]\tvalidation_0-rmse:48402.1\n",
      "[162]\tvalidation_0-rmse:47760.5\n",
      "[164]\tvalidation_0-rmse:47148.7\n",
      "[166]\tvalidation_0-rmse:46532.8\n",
      "[168]\tvalidation_0-rmse:45917.1\n",
      "[170]\tvalidation_0-rmse:45336.7\n",
      "[172]\tvalidation_0-rmse:44723.1\n",
      "[174]\tvalidation_0-rmse:44157.8\n",
      "[176]\tvalidation_0-rmse:43586.5\n",
      "[178]\tvalidation_0-rmse:43053.7\n",
      "[180]\tvalidation_0-rmse:42505.3\n",
      "[182]\tvalidation_0-rmse:42007.2\n",
      "[184]\tvalidation_0-rmse:41518.6\n",
      "[186]\tvalidation_0-rmse:41019.9\n",
      "[188]\tvalidation_0-rmse:40541.4\n",
      "[190]\tvalidation_0-rmse:40070.6\n",
      "[192]\tvalidation_0-rmse:39587.2\n",
      "[194]\tvalidation_0-rmse:39116.7\n",
      "[196]\tvalidation_0-rmse:38665.1\n",
      "[198]\tvalidation_0-rmse:38246.6\n",
      "[200]\tvalidation_0-rmse:37852.8\n",
      "[202]\tvalidation_0-rmse:37425.4\n",
      "[204]\tvalidation_0-rmse:37046.4\n",
      "[206]\tvalidation_0-rmse:36645.7\n",
      "[208]\tvalidation_0-rmse:36252.8\n",
      "[210]\tvalidation_0-rmse:35886.7\n",
      "[212]\tvalidation_0-rmse:35537.8\n",
      "[214]\tvalidation_0-rmse:35191.1\n",
      "[216]\tvalidation_0-rmse:34861.9\n",
      "[218]\tvalidation_0-rmse:34541.6\n",
      "[220]\tvalidation_0-rmse:34245.6\n",
      "[222]\tvalidation_0-rmse:33955.3\n",
      "[224]\tvalidation_0-rmse:33655.4\n",
      "[226]\tvalidation_0-rmse:33351.3\n",
      "[228]\tvalidation_0-rmse:33073.7\n",
      "[230]\tvalidation_0-rmse:32805.5\n",
      "[232]\tvalidation_0-rmse:32537.7\n",
      "[234]\tvalidation_0-rmse:32279.3\n",
      "[236]\tvalidation_0-rmse:32029.1\n",
      "[238]\tvalidation_0-rmse:31786.2\n",
      "[240]\tvalidation_0-rmse:31536\n",
      "[242]\tvalidation_0-rmse:31317.8\n",
      "[244]\tvalidation_0-rmse:31102.6\n",
      "[246]\tvalidation_0-rmse:30888\n",
      "[248]\tvalidation_0-rmse:30688.7\n",
      "[250]\tvalidation_0-rmse:30491.2\n",
      "[252]\tvalidation_0-rmse:30296.3\n",
      "[254]\tvalidation_0-rmse:30112.1\n",
      "[256]\tvalidation_0-rmse:29937.5\n",
      "[258]\tvalidation_0-rmse:29771.9\n",
      "[260]\tvalidation_0-rmse:29603.3\n",
      "[262]\tvalidation_0-rmse:29440.4\n",
      "[264]\tvalidation_0-rmse:29280.8\n",
      "[266]\tvalidation_0-rmse:29128.1\n",
      "[268]\tvalidation_0-rmse:28987.6\n",
      "[270]\tvalidation_0-rmse:28852.8\n",
      "[272]\tvalidation_0-rmse:28721.8\n",
      "[274]\tvalidation_0-rmse:28591.8\n",
      "[276]\tvalidation_0-rmse:28459.5\n",
      "[278]\tvalidation_0-rmse:28341.4\n",
      "[280]\tvalidation_0-rmse:28228.2\n",
      "[282]\tvalidation_0-rmse:28102.2\n",
      "[284]\tvalidation_0-rmse:27985\n",
      "[286]\tvalidation_0-rmse:27879.5\n",
      "[288]\tvalidation_0-rmse:27768.2\n",
      "[290]\tvalidation_0-rmse:27656\n",
      "[292]\tvalidation_0-rmse:27560.3\n",
      "[294]\tvalidation_0-rmse:27443.5\n",
      "[296]\tvalidation_0-rmse:27339.2\n",
      "[298]\tvalidation_0-rmse:27240.5\n",
      "[300]\tvalidation_0-rmse:27141.5\n",
      "[302]\tvalidation_0-rmse:27044.9\n",
      "[304]\tvalidation_0-rmse:26950.3\n",
      "[306]\tvalidation_0-rmse:26863.1\n",
      "[308]\tvalidation_0-rmse:26777.7\n",
      "[310]\tvalidation_0-rmse:26690.6\n",
      "[312]\tvalidation_0-rmse:26611.9\n",
      "[314]\tvalidation_0-rmse:26523.6\n",
      "[316]\tvalidation_0-rmse:26448.2\n",
      "[318]\tvalidation_0-rmse:26367.6\n",
      "[320]\tvalidation_0-rmse:26291.5\n",
      "[322]\tvalidation_0-rmse:26225.1\n",
      "[324]\tvalidation_0-rmse:26152.2\n",
      "[326]\tvalidation_0-rmse:26080.3\n",
      "[328]\tvalidation_0-rmse:26004\n",
      "[330]\tvalidation_0-rmse:25938.6\n",
      "[332]\tvalidation_0-rmse:25865.3\n",
      "[334]\tvalidation_0-rmse:25801.2\n",
      "[336]\tvalidation_0-rmse:25741.1\n",
      "[338]\tvalidation_0-rmse:25672.5\n",
      "[340]\tvalidation_0-rmse:25618\n",
      "[342]\tvalidation_0-rmse:25559.2\n",
      "[344]\tvalidation_0-rmse:25509.7\n",
      "[346]\tvalidation_0-rmse:25447.9\n",
      "[348]\tvalidation_0-rmse:25400\n",
      "[350]\tvalidation_0-rmse:25347.1\n",
      "[352]\tvalidation_0-rmse:25291.3\n",
      "[354]\tvalidation_0-rmse:25237.5\n",
      "[356]\tvalidation_0-rmse:25193.4\n",
      "[358]\tvalidation_0-rmse:25150.2\n",
      "[360]\tvalidation_0-rmse:25106.9\n",
      "[362]\tvalidation_0-rmse:25060.7\n",
      "[364]\tvalidation_0-rmse:25022.8\n",
      "[366]\tvalidation_0-rmse:24975.8\n",
      "[368]\tvalidation_0-rmse:24936.2\n",
      "[370]\tvalidation_0-rmse:24902.5\n",
      "[372]\tvalidation_0-rmse:24862.9\n",
      "[374]\tvalidation_0-rmse:24832.9\n",
      "[376]\tvalidation_0-rmse:24798.2\n",
      "[378]\tvalidation_0-rmse:24765.2\n",
      "[380]\tvalidation_0-rmse:24736.7\n",
      "[382]\tvalidation_0-rmse:24709.8\n",
      "[384]\tvalidation_0-rmse:24680.8\n",
      "[386]\tvalidation_0-rmse:24647\n",
      "[388]\tvalidation_0-rmse:24609.7\n",
      "[390]\tvalidation_0-rmse:24576.8\n",
      "[392]\tvalidation_0-rmse:24551.2\n",
      "[394]\tvalidation_0-rmse:24520.7\n",
      "[396]\tvalidation_0-rmse:24492.5\n",
      "[398]\tvalidation_0-rmse:24468\n",
      "[400]\tvalidation_0-rmse:24438.8\n",
      "[402]\tvalidation_0-rmse:24409.4\n",
      "[404]\tvalidation_0-rmse:24385.8\n",
      "[406]\tvalidation_0-rmse:24361.7\n",
      "[408]\tvalidation_0-rmse:24333.3\n",
      "[410]\tvalidation_0-rmse:24306.1\n",
      "[412]\tvalidation_0-rmse:24288.2\n",
      "[414]\tvalidation_0-rmse:24266.4\n",
      "[416]\tvalidation_0-rmse:24244.3\n",
      "[418]\tvalidation_0-rmse:24222.7\n",
      "[420]\tvalidation_0-rmse:24198.7\n",
      "[422]\tvalidation_0-rmse:24180.5\n",
      "[424]\tvalidation_0-rmse:24157.2\n",
      "[426]\tvalidation_0-rmse:24140\n",
      "[428]\tvalidation_0-rmse:24119\n",
      "[430]\tvalidation_0-rmse:24101.9\n",
      "[432]\tvalidation_0-rmse:24083.2\n",
      "[434]\tvalidation_0-rmse:24066.4\n",
      "[436]\tvalidation_0-rmse:24052.7\n",
      "[438]\tvalidation_0-rmse:24036.9\n",
      "[440]\tvalidation_0-rmse:24018.4\n",
      "[442]\tvalidation_0-rmse:23995\n",
      "[444]\tvalidation_0-rmse:23980.8\n",
      "[446]\tvalidation_0-rmse:23967.1\n",
      "[448]\tvalidation_0-rmse:23949.3\n",
      "[450]\tvalidation_0-rmse:23932.6\n",
      "[452]\tvalidation_0-rmse:23918.6\n",
      "[454]\tvalidation_0-rmse:23905.3\n",
      "[456]\tvalidation_0-rmse:23892.5\n",
      "[458]\tvalidation_0-rmse:23884.4\n",
      "[460]\tvalidation_0-rmse:23874.1\n",
      "[462]\tvalidation_0-rmse:23862.5\n",
      "[464]\tvalidation_0-rmse:23859.2\n",
      "[466]\tvalidation_0-rmse:23845.1\n",
      "[468]\tvalidation_0-rmse:23836.6\n",
      "[470]\tvalidation_0-rmse:23829.5\n",
      "[472]\tvalidation_0-rmse:23826.1\n",
      "[474]\tvalidation_0-rmse:23818.6\n",
      "[476]\tvalidation_0-rmse:23810.3\n",
      "[478]\tvalidation_0-rmse:23801.4\n",
      "[480]\tvalidation_0-rmse:23797.8\n",
      "[482]\tvalidation_0-rmse:23792.7\n",
      "[484]\tvalidation_0-rmse:23784.5\n",
      "[486]\tvalidation_0-rmse:23778.8\n",
      "[488]\tvalidation_0-rmse:23773.5\n",
      "[490]\tvalidation_0-rmse:23769.5\n",
      "[492]\tvalidation_0-rmse:23762.5\n",
      "[494]\tvalidation_0-rmse:23758.3\n",
      "[496]\tvalidation_0-rmse:23755.6\n",
      "[498]\tvalidation_0-rmse:23750.9\n",
      "[500]\tvalidation_0-rmse:23746.8\n",
      "[502]\tvalidation_0-rmse:23743.7\n",
      "[504]\tvalidation_0-rmse:23740.4\n",
      "[506]\tvalidation_0-rmse:23738.3\n",
      "[508]\tvalidation_0-rmse:23734.8\n",
      "[510]\tvalidation_0-rmse:23732.3\n",
      "[512]\tvalidation_0-rmse:23727.6\n",
      "[514]\tvalidation_0-rmse:23724.7\n"
     ]
    },
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "[516]\tvalidation_0-rmse:23710.2\n",
      "[518]\tvalidation_0-rmse:23708.9\n",
      "[520]\tvalidation_0-rmse:23703\n",
      "[522]\tvalidation_0-rmse:23693.1\n",
      "[524]\tvalidation_0-rmse:23691.9\n",
      "[526]\tvalidation_0-rmse:23685.1\n",
      "[528]\tvalidation_0-rmse:23677.8\n",
      "[530]\tvalidation_0-rmse:23673\n",
      "[532]\tvalidation_0-rmse:23658.8\n",
      "[534]\tvalidation_0-rmse:23652.2\n",
      "[536]\tvalidation_0-rmse:23642.2\n",
      "[538]\tvalidation_0-rmse:23641.9\n",
      "[540]\tvalidation_0-rmse:23625.2\n",
      "[542]\tvalidation_0-rmse:23626.7\n",
      "[544]\tvalidation_0-rmse:23618.8\n",
      "[546]\tvalidation_0-rmse:23620.6\n",
      "[548]\tvalidation_0-rmse:23617.6\n",
      "[550]\tvalidation_0-rmse:23601.9\n",
      "[552]\tvalidation_0-rmse:23595.8\n",
      "[554]\tvalidation_0-rmse:23593.9\n",
      "[556]\tvalidation_0-rmse:23591.3\n",
      "[558]\tvalidation_0-rmse:23587.9\n",
      "[560]\tvalidation_0-rmse:23581.4\n",
      "[562]\tvalidation_0-rmse:23574.4\n",
      "[564]\tvalidation_0-rmse:23574.3\n",
      "[566]\tvalidation_0-rmse:23569.2\n",
      "[568]\tvalidation_0-rmse:23558.3\n",
      "[570]\tvalidation_0-rmse:23554.5\n",
      "[572]\tvalidation_0-rmse:23550.1\n",
      "[574]\tvalidation_0-rmse:23539.2\n",
      "[576]\tvalidation_0-rmse:23533.8\n",
      "[578]\tvalidation_0-rmse:23534\n",
      "[580]\tvalidation_0-rmse:23531.1\n",
      "[582]\tvalidation_0-rmse:23525.7\n",
      "[584]\tvalidation_0-rmse:23518.9\n",
      "[586]\tvalidation_0-rmse:23515\n",
      "[588]\tvalidation_0-rmse:23512.4\n",
      "[590]\tvalidation_0-rmse:23512.3\n",
      "[592]\tvalidation_0-rmse:23508.3\n",
      "[594]\tvalidation_0-rmse:23503.3\n",
      "[596]\tvalidation_0-rmse:23498.3\n",
      "[598]\tvalidation_0-rmse:23491.5\n",
      "[600]\tvalidation_0-rmse:23485.7\n",
      "[602]\tvalidation_0-rmse:23481.3\n",
      "[604]\tvalidation_0-rmse:23478.3\n",
      "[606]\tvalidation_0-rmse:23470.5\n",
      "[608]\tvalidation_0-rmse:23466.3\n",
      "[610]\tvalidation_0-rmse:23463.5\n",
      "[612]\tvalidation_0-rmse:23460.2\n",
      "[614]\tvalidation_0-rmse:23457.2\n",
      "[616]\tvalidation_0-rmse:23453.4\n",
      "[618]\tvalidation_0-rmse:23449.1\n",
      "[620]\tvalidation_0-rmse:23444\n",
      "[622]\tvalidation_0-rmse:23440\n",
      "[624]\tvalidation_0-rmse:23436\n",
      "[626]\tvalidation_0-rmse:23432.7\n",
      "[628]\tvalidation_0-rmse:23429.6\n",
      "[630]\tvalidation_0-rmse:23425.1\n",
      "[632]\tvalidation_0-rmse:23425.3\n",
      "[634]\tvalidation_0-rmse:23419.5\n",
      "[636]\tvalidation_0-rmse:23416.5\n",
      "[638]\tvalidation_0-rmse:23412.3\n",
      "[640]\tvalidation_0-rmse:23413.1\n",
      "[642]\tvalidation_0-rmse:23408.3\n",
      "[644]\tvalidation_0-rmse:23406\n",
      "[646]\tvalidation_0-rmse:23403.3\n",
      "[648]\tvalidation_0-rmse:23395.8\n",
      "[650]\tvalidation_0-rmse:23394.9\n",
      "[652]\tvalidation_0-rmse:23390.6\n",
      "[654]\tvalidation_0-rmse:23388.5\n",
      "[656]\tvalidation_0-rmse:23381.8\n",
      "[658]\tvalidation_0-rmse:23379.3\n",
      "[660]\tvalidation_0-rmse:23380.4\n",
      "[662]\tvalidation_0-rmse:23380.4\n",
      "[664]\tvalidation_0-rmse:23376.8\n",
      "[666]\tvalidation_0-rmse:23370.4\n",
      "[668]\tvalidation_0-rmse:23369.7\n",
      "[670]\tvalidation_0-rmse:23368.1\n",
      "[672]\tvalidation_0-rmse:23368.5\n",
      "[674]\tvalidation_0-rmse:23366.3\n",
      "[676]\tvalidation_0-rmse:23364.3\n",
      "[678]\tvalidation_0-rmse:23360.1\n",
      "[680]\tvalidation_0-rmse:23358.4\n",
      "[682]\tvalidation_0-rmse:23351.7\n",
      "[684]\tvalidation_0-rmse:23347.9\n",
      "[686]\tvalidation_0-rmse:23344.4\n",
      "[688]\tvalidation_0-rmse:23341.9\n",
      "[690]\tvalidation_0-rmse:23341\n",
      "[692]\tvalidation_0-rmse:23341.8\n",
      "[694]\tvalidation_0-rmse:23340.3\n",
      "[696]\tvalidation_0-rmse:23335.7\n",
      "[698]\tvalidation_0-rmse:23327.3\n",
      "[700]\tvalidation_0-rmse:23330.2\n",
      "[702]\tvalidation_0-rmse:23327.5\n",
      "[704]\tvalidation_0-rmse:23326.4\n",
      "[706]\tvalidation_0-rmse:23324.9\n",
      "[708]\tvalidation_0-rmse:23327.1\n",
      "[710]\tvalidation_0-rmse:23326\n",
      "[712]\tvalidation_0-rmse:23325.2\n",
      "[714]\tvalidation_0-rmse:23323.8\n",
      "[716]\tvalidation_0-rmse:23324.3\n",
      "[718]\tvalidation_0-rmse:23320.6\n",
      "[720]\tvalidation_0-rmse:23317.8\n",
      "[722]\tvalidation_0-rmse:23315.5\n",
      "[724]\tvalidation_0-rmse:23317.6\n",
      "[726]\tvalidation_0-rmse:23318.7\n",
      "[728]\tvalidation_0-rmse:23316.3\n",
      "[730]\tvalidation_0-rmse:23315.3\n",
      "[732]\tvalidation_0-rmse:23314.1\n",
      "[734]\tvalidation_0-rmse:23310.3\n",
      "[736]\tvalidation_0-rmse:23307.7\n",
      "[738]\tvalidation_0-rmse:23307.2\n",
      "[740]\tvalidation_0-rmse:23302.8\n",
      "[742]\tvalidation_0-rmse:23298.8\n",
      "[744]\tvalidation_0-rmse:23297\n",
      "[746]\tvalidation_0-rmse:23295.1\n",
      "[748]\tvalidation_0-rmse:23295.2\n",
      "[750]\tvalidation_0-rmse:23294.5\n",
      "[752]\tvalidation_0-rmse:23295.4\n",
      "[754]\tvalidation_0-rmse:23295.6\n",
      "[756]\tvalidation_0-rmse:23291.3\n",
      "[758]\tvalidation_0-rmse:23288.3\n",
      "[760]\tvalidation_0-rmse:23286\n",
      "[762]\tvalidation_0-rmse:23284.4\n",
      "[764]\tvalidation_0-rmse:23287.5\n",
      "[766]\tvalidation_0-rmse:23285.2\n",
      "[768]\tvalidation_0-rmse:23281.5\n",
      "[770]\tvalidation_0-rmse:23282.3\n",
      "[772]\tvalidation_0-rmse:23281.7\n",
      "[774]\tvalidation_0-rmse:23282.6\n",
      "[776]\tvalidation_0-rmse:23280.3\n",
      "[778]\tvalidation_0-rmse:23278.5\n",
      "[780]\tvalidation_0-rmse:23278.7\n",
      "[782]\tvalidation_0-rmse:23277\n",
      "[784]\tvalidation_0-rmse:23275.1\n",
      "[786]\tvalidation_0-rmse:23276.2\n",
      "[788]\tvalidation_0-rmse:23278.1\n",
      "[790]\tvalidation_0-rmse:23277.1\n",
      "[792]\tvalidation_0-rmse:23276.7\n",
      "[794]\tvalidation_0-rmse:23274.7\n",
      "[796]\tvalidation_0-rmse:23276.9\n",
      "[798]\tvalidation_0-rmse:23275.9\n",
      "[800]\tvalidation_0-rmse:23276.1\n",
      "[802]\tvalidation_0-rmse:23277.1\n",
      "[804]\tvalidation_0-rmse:23275.1\n",
      "[806]\tvalidation_0-rmse:23273.5\n",
      "[808]\tvalidation_0-rmse:23273.2\n",
      "[810]\tvalidation_0-rmse:23273.7\n",
      "[812]\tvalidation_0-rmse:23275.4\n",
      "[814]\tvalidation_0-rmse:23271.6\n",
      "[816]\tvalidation_0-rmse:23271.4\n",
      "[818]\tvalidation_0-rmse:23271.1\n",
      "[820]\tvalidation_0-rmse:23271.9\n",
      "[822]\tvalidation_0-rmse:23271.9\n",
      "[824]\tvalidation_0-rmse:23271.6\n",
      "[826]\tvalidation_0-rmse:23269.2\n",
      "[828]\tvalidation_0-rmse:23266.6\n",
      "[830]\tvalidation_0-rmse:23267\n",
      "[832]\tvalidation_0-rmse:23266.9\n",
      "[834]\tvalidation_0-rmse:23268.1\n",
      "[836]\tvalidation_0-rmse:23267.7\n",
      "[838]\tvalidation_0-rmse:23268.1\n",
      "[840]\tvalidation_0-rmse:23270.1\n",
      "[842]\tvalidation_0-rmse:23269.9\n",
      "[844]\tvalidation_0-rmse:23269.6\n",
      "[846]\tvalidation_0-rmse:23270.1\n",
      "[848]\tvalidation_0-rmse:23270.6\n",
      "[850]\tvalidation_0-rmse:23271.6\n",
      "[852]\tvalidation_0-rmse:23267.1\n",
      "[854]\tvalidation_0-rmse:23266.8\n",
      "[856]\tvalidation_0-rmse:23266.7\n",
      "[858]\tvalidation_0-rmse:23264.4\n",
      "[860]\tvalidation_0-rmse:23262.7\n",
      "[862]\tvalidation_0-rmse:23258.9\n",
      "[864]\tvalidation_0-rmse:23258.7\n",
      "[866]\tvalidation_0-rmse:23256.3\n",
      "[868]\tvalidation_0-rmse:23257.7\n",
      "[870]\tvalidation_0-rmse:23259.3\n",
      "[872]\tvalidation_0-rmse:23257.7\n",
      "[874]\tvalidation_0-rmse:23254.8\n",
      "[876]\tvalidation_0-rmse:23254.7\n",
      "[878]\tvalidation_0-rmse:23251.5\n",
      "[880]\tvalidation_0-rmse:23249.4\n",
      "[882]\tvalidation_0-rmse:23248.9\n",
      "[884]\tvalidation_0-rmse:23247.1\n",
      "[886]\tvalidation_0-rmse:23246.8\n",
      "[888]\tvalidation_0-rmse:23248.2\n",
      "[890]\tvalidation_0-rmse:23246.5\n",
      "[892]\tvalidation_0-rmse:23243\n",
      "[894]\tvalidation_0-rmse:23246.1\n",
      "[896]\tvalidation_0-rmse:23243.7\n",
      "[898]\tvalidation_0-rmse:23245.6\n",
      "[900]\tvalidation_0-rmse:23244\n",
      "[902]\tvalidation_0-rmse:23241.4\n",
      "[904]\tvalidation_0-rmse:23240.5\n",
      "[906]\tvalidation_0-rmse:23240.8\n",
      "[908]\tvalidation_0-rmse:23238.9\n",
      "[910]\tvalidation_0-rmse:23238.2\n",
      "[912]\tvalidation_0-rmse:23239\n",
      "[914]\tvalidation_0-rmse:23241.2\n",
      "[916]\tvalidation_0-rmse:23238.2\n",
      "[918]\tvalidation_0-rmse:23236.3\n",
      "[920]\tvalidation_0-rmse:23231.4\n",
      "[922]\tvalidation_0-rmse:23233.1\n",
      "[924]\tvalidation_0-rmse:23235.5\n",
      "[926]\tvalidation_0-rmse:23231.4\n",
      "[928]\tvalidation_0-rmse:23228\n",
      "[930]\tvalidation_0-rmse:23225.8\n",
      "[932]\tvalidation_0-rmse:23223.8\n",
      "[934]\tvalidation_0-rmse:23221.2\n",
      "[936]\tvalidation_0-rmse:23222.8\n",
      "[938]\tvalidation_0-rmse:23220.8\n",
      "[940]\tvalidation_0-rmse:23220.8\n",
      "[942]\tvalidation_0-rmse:23222.1\n",
      "[944]\tvalidation_0-rmse:23221.3\n",
      "[946]\tvalidation_0-rmse:23219.3\n",
      "[948]\tvalidation_0-rmse:23215.8\n",
      "[950]\tvalidation_0-rmse:23215.1\n",
      "[952]\tvalidation_0-rmse:23215.2\n",
      "[954]\tvalidation_0-rmse:23215.3\n",
      "[956]\tvalidation_0-rmse:23214.3\n",
      "[958]\tvalidation_0-rmse:23215.5\n",
      "[960]\tvalidation_0-rmse:23214.6\n",
      "[962]\tvalidation_0-rmse:23212.7\n",
      "[964]\tvalidation_0-rmse:23213.8\n",
      "[966]\tvalidation_0-rmse:23211.8\n",
      "[968]\tvalidation_0-rmse:23212.7\n",
      "[970]\tvalidation_0-rmse:23212\n",
      "[972]\tvalidation_0-rmse:23214.2\n",
      "[974]\tvalidation_0-rmse:23216.1\n",
      "[976]\tvalidation_0-rmse:23216.9\n",
      "[978]\tvalidation_0-rmse:23218.3\n",
      "[980]\tvalidation_0-rmse:23216.6\n",
      "[982]\tvalidation_0-rmse:23219.3\n",
      "[984]\tvalidation_0-rmse:23219.8\n",
      "[986]\tvalidation_0-rmse:23218.5\n",
      "[988]\tvalidation_0-rmse:23220.2\n",
      "[990]\tvalidation_0-rmse:23223.8\n",
      "[992]\tvalidation_0-rmse:23220.9\n",
      "[994]\tvalidation_0-rmse:23223.4\n",
      "[996]\tvalidation_0-rmse:23223.6\n",
      "[998]\tvalidation_0-rmse:23222.9\n",
      "[999]\tvalidation_0-rmse:23222.8\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "XGBRegressor(base_score=0.5, booster='gbtree', colsample_bylevel=1,\n",
       "       colsample_bytree=1, gamma=0, learning_rate=0.01, max_delta_step=0,\n",
       "       max_depth=5, min_child_weight=1, missing=None, n_estimators=1000,\n",
       "       n_jobs=1, nthread=None, objective='reg:linear', random_state=0,\n",
       "       reg_alpha=0, reg_lambda=1, scale_pos_weight=1, seed=None,\n",
       "       silent=True, subsample=1)"
      ]
     },
     "execution_count": 107,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "model = XGBRegressor(max_depth=5,learning_rate=1e-2,n_estimators=1000)\n",
    "model.fit(x_train,y_train,\\\n",
    "          eval_metric='rmse',\\\n",
    "          eval_set=[(x_val,y_val)],\\\n",
    "          early_stopping_rounds=200,\\\n",
    "          verbose=2)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 108,
   "metadata": {},
   "outputs": [],
   "source": [
    "pred = model.predict(test_xgb_select)\n",
    "submission = pd.read_csv('sample_submission.csv')\n",
    "submission.SalePrice = pred\n",
    "submission.to_csv('submission_xgb_select.csv',index=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "### Ensemble"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 109,
   "metadata": {},
   "outputs": [],
   "source": [
    "pred1 = pd.read_csv('submission_xgb_select.csv')\n",
    "pred2 = pd.read_csv('submission_l1_select.csv')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 110,
   "metadata": {},
   "outputs": [],
   "source": [
    "pred = 0.7*pred1.SalePrice.values + 0.3*pred2.SalePrice.values"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 111,
   "metadata": {},
   "outputs": [],
   "source": [
    "submission = pd.read_csv('sample_submission.csv')\n",
    "submission.SalePrice = pred\n",
    "submission.to_csv('submission_ensemble.csv',index=False)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "AMMI",
   "language": "python",
   "name": "ammi"
  },
  "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
}