import pandas as pd
import numpy as np
from sklearn.model_selection import train_test_split, cross_val_score
from sklearn.metrics import mean_squared_error, r2_score
from sklearn.preprocessing import StandardScaler
import matplotlib.pyplot as plt
import lightgbm as lgb
import logging
LightGBMのログ出力をエラーのみ表示
logging.getLogger('lightgbm').setLevel(logging.ERROR)
データの読み込み
train_data = pd.read_csv('hour_train.csv')
test_data = pd.read_csv('hour_test.csv')
新しい特徴量の作成
train_data['hr_8_17_18'] = train_data['hr'].apply(lambda x: 1 if x in [8, 17, 18] else 0)
test_data['hr_8_17_18'] = test_data['hr'].apply(lambda x: 1 if x in [8, 17, 18] else 0)
train_data['atemp_0.2_0.8'] = train_data['atemp'].apply(lambda x: 1 if 0.2 <= x <= 0.8 else 0)
test_data['atemp_0.2_0.8'] = test_data['atemp'].apply(lambda x: 1 if 0.2 <= x <= 0.8 else 0)
train_data['temp_0.2_0.8'] = train_data['temp'].apply(lambda x: 1 if 0.2 <= x <= 0.8 else 0)
test_data['temp_0.2_0.8'] = test_data['temp'].apply(lambda x: 1 if 0.2 <= x <= 0.8 else 0)
train_data['hum_above_0.2'] = train_data['hum'].apply(lambda x: 1 if x >= 0.2 else 0)
test_data['hum_above_0.2'] = test_data['hum'].apply(lambda x: 1 if x >= 0.2 else 0)
train_data['windspeed_below_0.6'] = train_data['windspeed'].apply(lambda x: 1 if x <= 0.6 else 0)
test_data['windspeed_below_0.6'] = test_data['windspeed'].apply(lambda x: 1 if x <= 0.6 else 0)
train_data['hr_7_20'] = train_data['hr'].apply(lambda x: 1 if 7 <= x <= 20 else 0)
test_data['hr_7_20'] = test_data['hr'].apply(lambda x: 1 if 7 <= x <= 20 else 0)
'season'と'weathersit'をダミー変数に変換
train_data = pd.get_dummies(train_data, columns=['season', 'weathersit'], drop_first=True)
test_data = pd.get_dummies(test_data, columns=['season', 'weathersit'], drop_first=True)
トレーニングデータとテストデータでのダミー変数の列を一致させる
train_columns = set(train_data.columns)
test_columns = set(test_data.columns)
missing_in_test = train_columns - test_columns
missing_in_train = test_columns - train_columns
for col in missing_in_test:
test_data[col] = 0
for col in missing_in_train:
train_data[col] = 0
列を揃えるために並べ替え
train_data = train_data.sort_index(axis=1)
test_data = test_data.sort_index(axis=1)
特徴量とターゲット変数の設定
features = ['temp', 'atemp', 'yr', 'mnth', 'hr', 'hum', 'holiday', 'workingday',
'hr_8_17_18', 'atemp_0.2_0.8', 'temp_0.2_0.8', 'hum_above_0.2', 'hr_7_20'] +
[col for col in train_data.columns if col.startswith('season_') or col.startswith('weathersit_')]
target = 'cnt'
X_train = train_data[features]
y_train = train_data[target]
データの標準化
scaler = StandardScaler()
X_train_scaled = scaler.fit_transform(X_train)
LightGBMモデルの作成
model = lgb.LGBMRegressor(verbose=-1, random_state=42)
クロスバリデーションによるモデルの評価
cv_scores = cross_val_score(model, X_train_scaled, y_train, cv=5, scoring='neg_mean_squared_error')
mean_cv_mse = -cv_scores.mean()
mean_cv_r2 = cross_val_score(model, X_train_scaled, y_train, cv=5, scoring='r2').mean()
print(f'Cross-Validation Mean Squared Error: {mean_cv_mse:.2f}')
print(f'Cross-Validation R^2 Score: {mean_cv_r2:.2f}')
データの分割（トレーニングとバリデーション用）
X_train_split, X_val_split, y_train_split, y_val_split = train_test_split(X_train_scaled, y_train, test_size=0.2, random_state=42)
モデルのトレーニング
model.fit(X_train_split, y_train_split)
バリデーションデータでの予測
y_val_pred = model.predict(X_val_split)
バリデーションデータでの評価
print(f'Validation Mean Squared Error: {mean_squared_error(y_val_split, y_val_pred):.2f}')
print(f'Validation R^2 Score: {r2_score(y_val_split, y_val_pred):.2f}')
テストデータでの予測
X_test = test_data[features]
X_test_scaled = scaler.transform(X_test)
y_test_pred = model.predict(X_test_scaled)
テストデータの実際の値
y_test_actual = test_data[target]
テストデータの評価
test_mse = mean_squared_error(y_test_actual, y_test_pred)
test_r2 = r2_score(y_test_actual, y_test_pred)
print(f'Test Mean Squared Error: {test_mse:.2f}')
print(f'Test R^2 Score: {test_r2:.2f}')
時系列ごとの実測値と予測値の折れ線グラフを作成
test_data['y_test_pred'] = y_test_pred
時間帯ごとの実測値と予測値を計算
hourly_avg_actual = test_data.groupby('hr')[target].mean()
hourly_avg_pred = test_data.groupby('hr')['y_test_pred'].mean()
時間帯ごとの予測値と実測値の折れ線グラフを作成
plt.figure(figsize=(12, 6))
plt.plot(hourly_avg_actual.index, hourly_avg_actual, label='Actual', marker='o')
plt.plot(hourly_avg_pred.index, hourly_avg_pred, label='Predicted', marker='o')
plt.xlabel('Hour of the Day')
plt.ylabel('Average Count')
plt.title('Average Actual and Predicted Count by Hour of the Day')
plt.legend()
plt.grid(True)
plt.show()
dtedayごとの実測値と予測値の平均を計算
test_data['dteday'] = pd.to_datetime(test_data['dteday']) # dtedayをdatetime型に変換
daily_avg_actual = test_data.groupby('dteday')[target].mean()
daily_avg_pred = test_data.groupby('dteday')['y_test_pred'].mean()
daily_workingday = test_data.groupby('dteday')['workingday'].first()
dtedayごとの実測値と予測値の折れ線グラフを作成
plt.figure(figsize=(14, 7))
plt.plot(daily_avg_actual.index, daily_avg_actual, label='Actual', marker='o')
plt.plot(daily_avg_pred.index, daily_avg_pred, label='Predicted', marker='o')
plt.xlabel('Date')
plt.ylabel('Average Count')
plt.title('Average Actual and Predicted Count by Date')
plt.legend()
plt.grid(True)
workingdayの情報を横軸に追加
for date, is_workingday in daily_workingday.items():
plt.axvline(x=date, color='r' if is_workingday else 'b', linestyle='--', alpha=0.5)
plt.xticks(rotation=45)
plt.show()
1時間ごとの実測値と予測値の差の平均を算出
test_data['error'] = test_data[target] - test_data['y_test_pred']
hourly_avg_error = test_data.groupby('hr')['error'].mean()
1日の実測値と予測値の差の平均を算出
daily_avg_error = test_data.groupby('dteday')['error'].mean()
差の平均を表示
print("Hourly Average Error:")
print(hourly_avg_error)
print("\nDaily Average Error:")
print(daily_avg_error)