bike-sharing-demand/bike-sharing-demand-ensemble-model.ipynb

{
 "cells": [
  {
   "cell_type": "code",
   "execution_count": 1,
   "metadata": {},
   "outputs": [],
   "source": [
    "import pandas as pd\n",
    "import numpy as np\n",
    "\n",
    "import matplotlib as mpl\n",
    "import matplotlib.pyplot as plt\n",
    "import seaborn as sns\n",
    "\n",
    "# 노트북 안에 그래프를 그리기 위해\n",
    "%matplotlib inline\n",
    "\n",
    "# 그래프에서 격자로 숫자 범위가 눈에 잘 띄도록 ggplot 스타일을 사용\n",
    "plt.style.use('ggplot')\n",
    "\n",
    "# 그래프에서 마이너스 폰트 깨지는 문제에 대한 대처\n",
    "mpl.rcParams['axes.unicode_minus'] = False\n",
    "\n",
    "import warnings\n",
    "warnings.filterwarnings('ignore')"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 2,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(10886, 12)"
      ]
     },
     "execution_count": 2,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train = pd.read_csv(\"data/train.csv\", parse_dates=[\"datetime\"])\n",
    "train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 3,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(6493, 9)"
      ]
     },
     "execution_count": 3,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test = pd.read_csv(\"data/test.csv\", parse_dates=[\"datetime\"])\n",
    "test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 4,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(10886, 16)"
      ]
     },
     "execution_count": 4,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "train[\"year\"] = train[\"datetime\"].dt.year\n",
    "train[\"month\"] = train[\"datetime\"].dt.month\n",
    "train[\"hour\"] = train[\"datetime\"].dt.hour\n",
    "train[\"dayofweek\"] = train[\"datetime\"].dt.dayofweek\n",
    "train.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 5,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "(6493, 13)"
      ]
     },
     "execution_count": 5,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "test[\"year\"] = test[\"datetime\"].dt.year\n",
    "test[\"month\"] = test[\"datetime\"].dt.month\n",
    "test[\"hour\"] = test[\"datetime\"].dt.hour\n",
    "test[\"dayofweek\"] = test[\"datetime\"].dt.dayofweek\n",
    "test.shape"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 6,
   "metadata": {},
   "outputs": [],
   "source": [
    "# 연속형 feature와 범주형 feature\n",
    "# 범주형 feature의 type을 category로 변경 해 준다.\n",
    "categorical_feature_names = [\"season\",\"holiday\",\"workingday\",\"weather\",\n",
    "                             \"dayofweek\",\"month\",\"year\",\"hour\"]"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 7,
   "metadata": {},
   "outputs": [],
   "source": [
    "for var in categorical_feature_names:\n",
    "    train[var] = train[var].astype(\"category\")\n",
    "    test[var] = test[var].astype(\"category\")"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 8,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "['season',\n",
       " 'weather',\n",
       " 'temp',\n",
       " 'atemp',\n",
       " 'humidity',\n",
       " 'year',\n",
       " 'hour',\n",
       " 'dayofweek',\n",
       " 'holiday',\n",
       " 'workingday']"
      ]
     },
     "execution_count": 8,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "feature_names = [\"season\", \"weather\", \"temp\", \"atemp\", \"humidity\",\n",
    "                 \"year\", \"hour\", \"dayofweek\", \"holiday\", \"workingday\"\n",
    "                ]\n",
    "\n",
    "feature_names"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 9,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(10886, 10)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>season</th>\n",
       "      <th>weather</th>\n",
       "      <th>temp</th>\n",
       "      <th>atemp</th>\n",
       "      <th>humidity</th>\n",
       "      <th>year</th>\n",
       "      <th>hour</th>\n",
       "      <th>dayofweek</th>\n",
       "      <th>holiday</th>\n",
       "      <th>workingday</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>9.84</td>\n",
       "      <td>14.395</td>\n",
       "      <td>81</td>\n",
       "      <td>2011</td>\n",
       "      <td>0</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>9.02</td>\n",
       "      <td>13.635</td>\n",
       "      <td>80</td>\n",
       "      <td>2011</td>\n",
       "      <td>1</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>9.02</td>\n",
       "      <td>13.635</td>\n",
       "      <td>80</td>\n",
       "      <td>2011</td>\n",
       "      <td>2</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>9.84</td>\n",
       "      <td>14.395</td>\n",
       "      <td>75</td>\n",
       "      <td>2011</td>\n",
       "      <td>3</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>9.84</td>\n",
       "      <td>14.395</td>\n",
       "      <td>75</td>\n",
       "      <td>2011</td>\n",
       "      <td>4</td>\n",
       "      <td>5</td>\n",
       "      <td>0</td>\n",
       "      <td>0</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  season weather  temp   atemp  humidity  year hour dayofweek holiday  \\\n",
       "0      1       1  9.84  14.395        81  2011    0         5       0   \n",
       "1      1       1  9.02  13.635        80  2011    1         5       0   \n",
       "2      1       1  9.02  13.635        80  2011    2         5       0   \n",
       "3      1       1  9.84  14.395        75  2011    3         5       0   \n",
       "4      1       1  9.84  14.395        75  2011    4         5       0   \n",
       "\n",
       "  workingday  \n",
       "0          0  \n",
       "1          0  \n",
       "2          0  \n",
       "3          0  \n",
       "4          0  "
      ]
     },
     "execution_count": 9,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_train = train[feature_names]\n",
    "\n",
    "print(X_train.shape)\n",
    "X_train.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 10,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(6493, 10)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>season</th>\n",
       "      <th>weather</th>\n",
       "      <th>temp</th>\n",
       "      <th>atemp</th>\n",
       "      <th>humidity</th>\n",
       "      <th>year</th>\n",
       "      <th>hour</th>\n",
       "      <th>dayofweek</th>\n",
       "      <th>holiday</th>\n",
       "      <th>workingday</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>10.66</td>\n",
       "      <td>11.365</td>\n",
       "      <td>56</td>\n",
       "      <td>2011</td>\n",
       "      <td>0</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>10.66</td>\n",
       "      <td>13.635</td>\n",
       "      <td>56</td>\n",
       "      <td>2011</td>\n",
       "      <td>1</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>10.66</td>\n",
       "      <td>13.635</td>\n",
       "      <td>56</td>\n",
       "      <td>2011</td>\n",
       "      <td>2</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>10.66</td>\n",
       "      <td>12.880</td>\n",
       "      <td>56</td>\n",
       "      <td>2011</td>\n",
       "      <td>3</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>1</td>\n",
       "      <td>1</td>\n",
       "      <td>10.66</td>\n",
       "      <td>12.880</td>\n",
       "      <td>56</td>\n",
       "      <td>2011</td>\n",
       "      <td>4</td>\n",
       "      <td>3</td>\n",
       "      <td>0</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "  season weather   temp   atemp  humidity  year hour dayofweek holiday  \\\n",
       "0      1       1  10.66  11.365        56  2011    0         3       0   \n",
       "1      1       1  10.66  13.635        56  2011    1         3       0   \n",
       "2      1       1  10.66  13.635        56  2011    2         3       0   \n",
       "3      1       1  10.66  12.880        56  2011    3         3       0   \n",
       "4      1       1  10.66  12.880        56  2011    4         3       0   \n",
       "\n",
       "  workingday  \n",
       "0          1  \n",
       "1          1  \n",
       "2          1  \n",
       "3          1  \n",
       "4          1  "
      ]
     },
     "execution_count": 10,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "X_test = test[feature_names]\n",
    "\n",
    "print(X_test.shape)\n",
    "X_test.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 11,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(10886,)\n"
     ]
    },
    {
     "data": {
      "text/plain": [
       "0    16\n",
       "1    40\n",
       "2    32\n",
       "3    13\n",
       "4     1\n",
       "Name: count, dtype: int64"
      ]
     },
     "execution_count": 11,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "label_name = \"count\"\n",
    "\n",
    "y_train = train[label_name]\n",
    "\n",
    "print(y_train.shape)\n",
    "y_train.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 12,
   "metadata": {},
   "outputs": [],
   "source": [
    "from sklearn.metrics import make_scorer\n",
    "\n",
    "def rmsle(predicted_values, actual_values, convertExp=True):\n",
    "\n",
    "    if convertExp:\n",
    "        predicted_values = np.exp(predicted_values),\n",
    "        actual_values = np.exp(actual_values)\n",
    "        \n",
    "    # 넘파이로 배열 형태로 바꿔준다.\n",
    "    predicted_values = np.array(predicted_values)\n",
    "    actual_values = np.array(actual_values)\n",
    "    \n",
    "    # 예측값과 실제 값에 1을 더하고 로그를 씌워준다.\n",
    "    # 값이 0일 수도 있어서 로그를 취했을 때 마이너스 무한대가 될 수도 있기 때문에 1을 더해 줌\n",
    "    # 로그를 씌워주는 것은 정규분포로 만들어주기 위해\n",
    "    log_predict = np.log(predicted_values + 1)\n",
    "    log_actual = np.log(actual_values + 1)\n",
    "    \n",
    "    # 위에서 계산한 예측값에서 실제값을 빼주고 제곱을 해준다.\n",
    "    difference = log_predict - log_actual\n",
    "    difference = np.square(difference)\n",
    "    \n",
    "    # 평균을 낸다.\n",
    "    mean_difference = difference.mean()\n",
    "    \n",
    "    # 다시 루트를 씌운다.\n",
    "    score = np.sqrt(mean_difference)\n",
    "    \n",
    "    return score"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 선형회귀 모델 Linear Regression Model\n",
    "* 선형회귀 또는 최소제곱법은 가장 간단하고 오래된 회귀용 선형 알고리즘\n",
    "* 선형회귀는 예측과 훈련 세트에 있는 타깃 y사이의 평균제곱오차(MSE)를 최소화하는 파라미터 w와 b를 찾는다.\n",
    "* 매개변수가 없는 것이 장점이지만, 모델의 복잡도를 제어할 수 없다는 단점이 있다.\n",
    "* 아래 위키피디아에서 가져 온 그래프에서 파란선이 선형회귀 모델을 나타내고 빨간점들이 훈련데이터를 나타낸다.\n",
    "\n",
    "![선형회귀 위키피디아 이미지](https://upload.wikimedia.org/wikipedia/commons/b/be/Normdist_regression.png)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 13,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "RMSLE Value For Linear Regression:  0.9803697923313522\n"
     ]
    }
   ],
   "source": [
    "from sklearn.linear_model import LinearRegression, Ridge, Lasso\n",
    "from sklearn.model_selection import GridSearchCV\n",
    "from sklearn import metrics\n",
    "import warnings\n",
    "pd.options.mode.chained_assignment = None\n",
    "warnings.filterwarnings(\"ignore\", category=DeprecationWarning)\n",
    "\n",
    "# 선형회귀 모델을 초기화\n",
    "lModel = LinearRegression()\n",
    "\n",
    "# 모델을 학습시킨다.\n",
    "y_train_log = np.log1p(y_train)\n",
    "lModel.fit(X_train, y_train_log)\n",
    "\n",
    "# 예측하고 정확도를 평가한다.\n",
    "preds = lModel.predict(X_train)\n",
    "print (\"RMSLE Value For Linear Regression: \",\n",
    "       rmsle(np.exp(y_train_log),np.exp(preds), False))"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 릿지 Regularization Model - Ridge\n",
    "* 회귀를 위한 선형모델\n",
    "* 가중치(w)의 모든 원소가 0에 가깝게 만들어 모든 피처가 주는 영향을 최소화(기울기를 작게 만듦)\n",
    "* Regularization(정규화)는 오버피팅(과대적합)이 되지 않도록 모델을 강제로 제한한다는 의미\n",
    "* max_iter(반복 실행하는 최대 횟수)는 3000을 넣어주었다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 14,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'alpha': 0.01, 'max_iter': 3000}\n",
      "RMSLE Value For Ridge Regression:  0.9803697902780835\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>mean_fit_time</th>\n",
       "      <th>std_fit_time</th>\n",
       "      <th>mean_score_time</th>\n",
       "      <th>std_score_time</th>\n",
       "      <th>param_alpha</th>\n",
       "      <th>param_max_iter</th>\n",
       "      <th>params</th>\n",
       "      <th>split0_test_score</th>\n",
       "      <th>split1_test_score</th>\n",
       "      <th>split2_test_score</th>\n",
       "      <th>split3_test_score</th>\n",
       "      <th>split4_test_score</th>\n",
       "      <th>mean_test_score</th>\n",
       "      <th>std_test_score</th>\n",
       "      <th>rank_test_score</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.012653</td>\n",
       "      <td>0.002890</td>\n",
       "      <td>0.004786</td>\n",
       "      <td>0.000813</td>\n",
       "      <td>0.01</td>\n",
       "      <td>3000</td>\n",
       "      <td>{'alpha': 0.01, 'max_iter': 3000}</td>\n",
       "      <td>-1.003138</td>\n",
       "      <td>-0.906726</td>\n",
       "      <td>-1.029928</td>\n",
       "      <td>-1.016631</td>\n",
       "      <td>-0.982313</td>\n",
       "      <td>-0.987749</td>\n",
       "      <td>0.043459</td>\n",
       "      <td>1</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.010568</td>\n",
       "      <td>0.000300</td>\n",
       "      <td>0.004176</td>\n",
       "      <td>0.000229</td>\n",
       "      <td>0.1</td>\n",
       "      <td>3000</td>\n",
       "      <td>{'alpha': 0.1, 'max_iter': 3000}</td>\n",
       "      <td>-1.003140</td>\n",
       "      <td>-0.906726</td>\n",
       "      <td>-1.029928</td>\n",
       "      <td>-1.016630</td>\n",
       "      <td>-0.982315</td>\n",
       "      <td>-0.987749</td>\n",
       "      <td>0.043459</td>\n",
       "      <td>2</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.011156</td>\n",
       "      <td>0.001189</td>\n",
       "      <td>0.004172</td>\n",
       "      <td>0.000109</td>\n",
       "      <td>1</td>\n",
       "      <td>3000</td>\n",
       "      <td>{'alpha': 1, 'max_iter': 3000}</td>\n",
       "      <td>-1.003167</td>\n",
       "      <td>-0.906727</td>\n",
       "      <td>-1.029927</td>\n",
       "      <td>-1.016621</td>\n",
       "      <td>-0.982331</td>\n",
       "      <td>-0.987756</td>\n",
       "      <td>0.043459</td>\n",
       "      <td>3</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.010919</td>\n",
       "      <td>0.000596</td>\n",
       "      <td>0.004402</td>\n",
       "      <td>0.000479</td>\n",
       "      <td>2</td>\n",
       "      <td>3000</td>\n",
       "      <td>{'alpha': 2, 'max_iter': 3000}</td>\n",
       "      <td>-1.003196</td>\n",
       "      <td>-0.906728</td>\n",
       "      <td>-1.029926</td>\n",
       "      <td>-1.016612</td>\n",
       "      <td>-0.982349</td>\n",
       "      <td>-0.987764</td>\n",
       "      <td>0.043459</td>\n",
       "      <td>4</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.011107</td>\n",
       "      <td>0.001719</td>\n",
       "      <td>0.004177</td>\n",
       "      <td>0.000158</td>\n",
       "      <td>3</td>\n",
       "      <td>3000</td>\n",
       "      <td>{'alpha': 3, 'max_iter': 3000}</td>\n",
       "      <td>-1.003226</td>\n",
       "      <td>-0.906729</td>\n",
       "      <td>-1.029925</td>\n",
       "      <td>-1.016602</td>\n",
       "      <td>-0.982366</td>\n",
       "      <td>-0.987771</td>\n",
       "      <td>0.043459</td>\n",
       "      <td>5</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   mean_fit_time  std_fit_time  mean_score_time  std_score_time param_alpha  \\\n",
       "0       0.012653      0.002890         0.004786        0.000813        0.01   \n",
       "1       0.010568      0.000300         0.004176        0.000229         0.1   \n",
       "2       0.011156      0.001189         0.004172        0.000109           1   \n",
       "3       0.010919      0.000596         0.004402        0.000479           2   \n",
       "4       0.011107      0.001719         0.004177        0.000158           3   \n",
       "\n",
       "  param_max_iter                             params  split0_test_score  \\\n",
       "0           3000  {'alpha': 0.01, 'max_iter': 3000}          -1.003138   \n",
       "1           3000   {'alpha': 0.1, 'max_iter': 3000}          -1.003140   \n",
       "2           3000     {'alpha': 1, 'max_iter': 3000}          -1.003167   \n",
       "3           3000     {'alpha': 2, 'max_iter': 3000}          -1.003196   \n",
       "4           3000     {'alpha': 3, 'max_iter': 3000}          -1.003226   \n",
       "\n",
       "   split1_test_score  split2_test_score  split3_test_score  split4_test_score  \\\n",
       "0          -0.906726          -1.029928          -1.016631          -0.982313   \n",
       "1          -0.906726          -1.029928          -1.016630          -0.982315   \n",
       "2          -0.906727          -1.029927          -1.016621          -0.982331   \n",
       "3          -0.906728          -1.029926          -1.016612          -0.982349   \n",
       "4          -0.906729          -1.029925          -1.016602          -0.982366   \n",
       "\n",
       "   mean_test_score  std_test_score  rank_test_score  \n",
       "0        -0.987749        0.043459                1  \n",
       "1        -0.987749        0.043459                2  \n",
       "2        -0.987756        0.043459                3  \n",
       "3        -0.987764        0.043459                4  \n",
       "4        -0.987771        0.043459                5  "
      ]
     },
     "execution_count": 14,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "ridge_m_ = Ridge()\n",
    "ridge_params_ = { 'max_iter':[3000],'alpha':[0.01, 0.1, 1, 2, 3, 4, 10, 30,100,200,300,400,800,900,1000]}\n",
    "rmsle_scorer = metrics.make_scorer(rmsle, greater_is_better=False)\n",
    "grid_ridge_m = GridSearchCV( ridge_m_,\n",
    "                          ridge_params_,\n",
    "                          scoring = rmsle_scorer,\n",
    "                          cv=5)\n",
    "\n",
    "y_train_log = np.log1p(y_train)\n",
    "grid_ridge_m.fit( X_train, y_train_log )\n",
    "preds = grid_ridge_m.predict(X_train)\n",
    "print (grid_ridge_m.best_params_)\n",
    "print (\"RMSLE Value For Ridge Regression: \", rmsle(np.exp(y_train_log),np.exp(preds), False))\n",
    "\n",
    "df = pd.DataFrame(grid_ridge_m.cv_results_)\n",
    "df.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 15,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>alpha</th>\n",
       "      <th>rmsle</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>0.01</td>\n",
       "      <td>0.987749</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>0.10</td>\n",
       "      <td>0.987749</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>1.00</td>\n",
       "      <td>0.987756</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2.00</td>\n",
       "      <td>0.987764</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>3.00</td>\n",
       "      <td>0.987771</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "   alpha     rmsle\n",
       "0   0.01  0.987749\n",
       "1   0.10  0.987749\n",
       "2   1.00  0.987756\n",
       "3   2.00  0.987764\n",
       "4   3.00  0.987771"
      ]
     },
     "execution_count": 15,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "df[\"alpha\"] = df[\"params\"].apply(lambda x:x[\"alpha\"])\n",
    "df[\"rmsle\"] = df[\"mean_test_score\"].apply(lambda x:-x)\n",
    "df[[\"alpha\", \"rmsle\"]].head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 16,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x12100e320>"
      ]
     },
     "execution_count": 16,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAuAAAAFQCAYAAAD+/MvhAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<Figure size 864x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig,ax= plt.subplots()\n",
    "fig.set_size_inches(12,5)\n",
    "plt.xticks(rotation=30, ha='right')\n",
    "sns.pointplot(data=df,x=\"alpha\",y=\"rmsle\",ax=ax)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 라쏘 Regularization Model - Lasso\n",
    "* 선형회귀의 Regularization(규제)를 적용하는 대안\n",
    "* 계수를 0에 가깝게 만들려고 하며 이를 L1규제라고 하며, 어떤 계수는 0이 되기도 하는데 이는 완전히 제외하는 피처가 생긴다는 의미다.\n",
    "* 피처 선택이 자동으로 이루어진다고도 볼 수 있다.\n",
    "* alpha 값의 기본 값은 1.0이며, 과소 적합을 줄이기 위해서는 이 값을 줄여야 한다.\n",
    "* 그리드 서치로 아래 라쏘모델을 실행했을 때 베스트 알파값은 0.0025\n",
    "* max_iter(반복 실행하는 최대 횟수)는 3000을 넣어주었다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 17,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "{'alpha': 0.00125, 'max_iter': 3000}\n",
      "RMSLE Value For Lasso Regression:  0.9803727821457646\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>alpha</th>\n",
       "      <th>rmsle</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>10.000000</td>\n",
       "      <td>1.397134</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>1.000000</td>\n",
       "      <td>1.071961</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>0.500000</td>\n",
       "      <td>1.062121</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>0.333333</td>\n",
       "      <td>1.060826</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>0.250000</td>\n",
       "      <td>1.061149</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "       alpha     rmsle\n",
       "0  10.000000  1.397134\n",
       "1   1.000000  1.071961\n",
       "2   0.500000  1.062121\n",
       "3   0.333333  1.060826\n",
       "4   0.250000  1.061149"
      ]
     },
     "execution_count": 17,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "lasso_m_ = Lasso()\n",
    "\n",
    "alpha  = 1/np.array([0.1, 1, 2, 3, 4, 10, 30,100,200,300,400,800,900,1000])\n",
    "lasso_params_ = { 'max_iter':[3000],'alpha':alpha}\n",
    "\n",
    "grid_lasso_m = GridSearchCV( lasso_m_,lasso_params_,scoring = rmsle_scorer,cv=5)\n",
    "y_train_log = np.log1p(y_train)\n",
    "grid_lasso_m.fit( X_train , y_train_log )\n",
    "preds = grid_lasso_m.predict(X_train)\n",
    "print (grid_lasso_m.best_params_)\n",
    "print (\"RMSLE Value For Lasso Regression: \",rmsle(np.exp(y_train_log),np.exp(preds),False))\n",
    "\n",
    "df = pd.DataFrame(grid_lasso_m.cv_results_)\n",
    "df[\"alpha\"] = df[\"params\"].apply(lambda x:x[\"alpha\"])\n",
    "df[\"rmsle\"] = df[\"mean_test_score\"].apply(lambda x:-x)\n",
    "\n",
    "df[[\"alpha\", \"rmsle\"]].head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 18,
   "metadata": {},
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x1211f8b38>"
      ]
     },
     "execution_count": 18,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAtoAAAGACAYAAACTE8U/AAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n7gAJtIUQQgghhLgDJNAWQgghhBDiDvg/Hw7uTjklZqYAAAAASUVORK5CYII=\n",
      "text/plain": [
       "<Figure size 864x360 with 1 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "fig,ax= plt.subplots()\n",
    "fig.set_size_inches(12,5)\n",
    "plt.xticks(rotation=30, ha='right')\n",
    "sns.pointplot(data=df,x=\"alpha\",y=\"rmsle\",ax=ax)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 앙상블모델 - 랜덤포레스트 \n",
    "### Ensemble Models - Random Forest"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 19,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "RMSLE Value For Random Forest:  0.1070944699920936\n"
     ]
    }
   ],
   "source": [
    "from sklearn.ensemble import RandomForestRegressor\n",
    "rfModel = RandomForestRegressor(n_estimators=100)\n",
    "\n",
    "y_train_log = np.log1p(y_train)\n",
    "rfModel.fit(X_train, y_train_log)\n",
    "\n",
    "preds = rfModel.predict(X_train)\n",
    "score = rmsle(np.exp(y_train_log),np.exp(preds),False)\n",
    "print (\"RMSLE Value For Random Forest: \",score)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## 앙상블모델 - 그라디언트 부스트\n",
    "### Ensemble Model - Gradient Boost\n",
    "* 여러개의 결정트리를 묶어 강력한 모델을 만드는 또 다른 앙상블 기법\n",
    "* 회귀와 분류에 모두 사용할 수 있음\n",
    "* 랜덤포레스트와 달리 이진 트리의 오차를 보완하는 방식으로 순차적으로 트리를 만든다.\n",
    "* 무작위성이 없고 강력한 사전 가지치기가 사용 됨\n",
    "* 1~5개의 깊지 않은 트리를 사용하기 때문에 메모리를 적게 사용하고 예측이 빠름\n",
    "* learning_rate : 오차를 얼마나 강하게 보정할 것인지를 제어\n",
    "* n_estimator의 값을 키우면 앙상블에 트리가 더 많이 추가 되어 모델의 복잡도가 커지고 훈련세트에서의 실수를 바로잡을 기회가 많아지지만 너무 크면 모델이 복잡해지고 오버피팅(과대적합)이 될 수있다.\n",
    "* max_depth(max_leaf_nodes) 복잡도를 너무 높이지 말고 트리의 깊이가 5보다 깊어지지 않게 한다."
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 20,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "RMSLE Value For Gradient Boost:  0.2135740372724937\n"
     ]
    }
   ],
   "source": [
    "from sklearn.ensemble import GradientBoostingRegressor\n",
    "gbm = GradientBoostingRegressor(n_estimators=4000, alpha=0.01);\n",
    "\n",
    "y_train_log = np.log1p(y_train)\n",
    "gbm.fit(X_train, y_train_log)\n",
    "\n",
    "preds = gbm.predict(X_train)\n",
    "score = rmsle(np.exp(y_train_log),np.exp(preds),False)\n",
    "print (\"RMSLE Value For Gradient Boost: \", score)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 21,
   "metadata": {
    "scrolled": true
   },
   "outputs": [
    {
     "data": {
      "text/plain": [
       "<matplotlib.axes._subplots.AxesSubplot at 0x12c950588>"
      ]
     },
     "execution_count": 21,
     "metadata": {},
     "output_type": "execute_result"
    },
    {
     "data": {
      "image/png": "iVBORw0KGgoAAAANSUhEUgAAAs8AAAFACAYAAABDfJEnAAAABHNCSVQICAgIfAhkiAAAAAlwSFlz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\n",
      "text/plain": [
       "<Figure size 864x360 with 2 Axes>"
      ]
     },
     "metadata": {},
     "output_type": "display_data"
    }
   ],
   "source": [
    "predsTest = gbm.predict(X_test)\n",
    "fig,(ax1,ax2)= plt.subplots(ncols=2)\n",
    "fig.set_size_inches(12,5)\n",
    "sns.distplot(y_train,ax=ax1,bins=50)\n",
    "sns.distplot(np.exp(predsTest),ax=ax2,bins=50)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "## Submit"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 22,
   "metadata": {},
   "outputs": [
    {
     "name": "stdout",
     "output_type": "stream",
     "text": [
      "(6493, 2)\n"
     ]
    },
    {
     "data": {
      "text/html": [
       "<div>\n",
       "<style scoped>\n",
       "    .dataframe tbody tr th:only-of-type {\n",
       "        vertical-align: middle;\n",
       "    }\n",
       "\n",
       "    .dataframe tbody tr th {\n",
       "        vertical-align: top;\n",
       "    }\n",
       "\n",
       "    .dataframe thead th {\n",
       "        text-align: right;\n",
       "    }\n",
       "</style>\n",
       "<table border=\"1\" class=\"dataframe\">\n",
       "  <thead>\n",
       "    <tr style=\"text-align: right;\">\n",
       "      <th></th>\n",
       "      <th>datetime</th>\n",
       "      <th>count</th>\n",
       "    </tr>\n",
       "  </thead>\n",
       "  <tbody>\n",
       "    <tr>\n",
       "      <th>0</th>\n",
       "      <td>2011-01-20 00:00:00</td>\n",
       "      <td>11.227543</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>1</th>\n",
       "      <td>2011-01-20 01:00:00</td>\n",
       "      <td>6.296610</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>2</th>\n",
       "      <td>2011-01-20 02:00:00</td>\n",
       "      <td>4.259238</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>3</th>\n",
       "      <td>2011-01-20 03:00:00</td>\n",
       "      <td>2.666615</td>\n",
       "    </tr>\n",
       "    <tr>\n",
       "      <th>4</th>\n",
       "      <td>2011-01-20 04:00:00</td>\n",
       "      <td>2.295918</td>\n",
       "    </tr>\n",
       "  </tbody>\n",
       "</table>\n",
       "</div>"
      ],
      "text/plain": [
       "              datetime      count\n",
       "0  2011-01-20 00:00:00  11.227543\n",
       "1  2011-01-20 01:00:00   6.296610\n",
       "2  2011-01-20 02:00:00   4.259238\n",
       "3  2011-01-20 03:00:00   2.666615\n",
       "4  2011-01-20 04:00:00   2.295918"
      ]
     },
     "execution_count": 22,
     "metadata": {},
     "output_type": "execute_result"
    }
   ],
   "source": [
    "submission = pd.read_csv(\"data/sampleSubmission.csv\")\n",
    "submission\n",
    "\n",
    "submission[\"count\"] = np.exp(predsTest)\n",
    "\n",
    "print(submission.shape)\n",
    "submission.head()"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": 23,
   "metadata": {},
   "outputs": [],
   "source": [
    "submission.to_csv(f\"data/Score_{score:.5f}_submission.csv\", index=False)"
   ]
  },
  {
   "cell_type": "markdown",
   "metadata": {},
   "source": [
    "참고 : \n",
    "* [EDA & Ensemble Model (Top 10 Percentile) | Kaggle](https://www.kaggle.com/viveksrinivasan/eda-ensemble-model-top-10-percentile)\n",
    "* [How to finish top 10 percentile in Bike Sharing Demand Competition In Kaggle? (part -1)](https://medium.com/@viveksrinivasan/how-to-finish-top-10-percentile-in-bike-sharing-demand-competition-in-kaggle-part-1-c816ea9c51e1)\n",
    "* [How to finish top 10 percentile in Bike Sharing Demand Competition In Kaggle? (part -2)](https://medium.com/@viveksrinivasan/how-to-finish-top-10-percentile-in-bike-sharing-demand-competition-in-kaggle-part-2-29e854aaab7d)\n",
    "* [Kaggle Bike Sharing Demand Prediction - How To Get In Top 5 Of Competition](https://www.analyticsvidhya.com/blog/2015/06/solution-kaggle-competition-bike-sharing-demand/)"
   ]
  },
  {
   "cell_type": "code",
   "execution_count": null,
   "metadata": {},
   "outputs": [],
   "source": []
  }
 ],
 "metadata": {
  "kernelspec": {
   "display_name": "Python 3",
   "language": "python",
   "name": "python3"
  },
  "language_info": {
   "codemirror_mode": {
    "name": "ipython",
    "version": 3
   },
   "file_extension": ".py",
   "mimetype": "text/x-python",
   "name": "python",
   "nbconvert_exporter": "python",
   "pygments_lexer": "ipython3",
   "version": "3.6.1"
  },
  "varInspector": {
   "cols": {
    "lenName": 16,
    "lenType": 16,
    "lenVar": 40
   },
   "kernels_config": {
    "python": {
     "delete_cmd_postfix": "",
     "delete_cmd_prefix": "del ",
     "library": "var_list.py",
     "varRefreshCmd": "print(var_dic_list())"
    },
    "r": {
     "delete_cmd_postfix": ") ",
     "delete_cmd_prefix": "rm(",
     "library": "var_list.r",
     "varRefreshCmd": "cat(var_dic_list()) "
    }
   },
   "types_to_exclude": [
    "module",
    "function",
    "builtin_function_or_method",
    "instance",
    "_Feature"
   ],
   "window_display": false
  }
 },
 "nbformat": 4,
 "nbformat_minor": 2
}