kaggle教程--9--交叉验证

交叉验证（Cross Validation）

 

Cross-Validation and Train-Test Split

数据很多的时候，用Train-Test Split，时间短

数据不多的时候，用Cross-Validation，模型的分数准

 

例子1：

import pandas as pd
data = pd.read_csv('../input/melb_data.csv')
cols_to_use = ['Rooms', 'Distance', 'Landsize', 'BuildingArea', 'YearBuilt']
X = data[cols_to_use]
y = data.Price

 

from sklearn.ensemble import RandomForestRegressor
from sklearn.pipeline import make_pipeline
from sklearn.preprocessing import Imputer
my_pipeline = make_pipeline(Imputer(), RandomForestRegressor())

 

from sklearn.model_selection import cross_val_score
scores = cross_val_score(my_pipeline, X, y, scoring='neg_mean_absolute_error')
print(scores)

 

print('Mean Absolute Error %2f' %(-1 * scores.mean()))

 

例子2 交叉验证+管道

from sklearn.model_selection import cross_val_score
my_pipeline = make_pipeline(Imputer(), XGBRegressor(n_estimators=1000, learning_rate=0.02))
scores = cross_val_score(my_pipeline, final_train_data, y, scoring='neg_mean_absolute_error')
print("交叉验证分数：")
print(scores)
print('Mean Absolute Error %2f' %(-1 * scores.mean()))

 

例子3：GridSearchCV

from sklearn.model_selection import GridSearchCV
# pipeline
XGB_pipeline = make_pipeline(SimpleImputer(), XGBRegressor())

# 要调参数
para_grid = {'xgbregressor__learning_rate': [0.01,0.1,0.5],'xgbregressor__n_estimators':[100,500,1000]}
grid = GridSearchCV(estimator=XGB_pipeline, param_grid = para_grid,cv=5)
grid.fit(final_train_data, y)

# 输出最佳参数组合
print (grid.best_params_)

注意：
1、要调参数为dict格式，参数名前需要加小写的XGBRegressor和双下划线也就是xgbregressor__
2、每次不要放太多参数，否则运行时间会很长，最好对所调参数有一定了解，甚至可以阅读相应已发表paper
3、GridSearchCV中的参数cv表示cross validation，cross validation对小数据量样本比适用，能充分利用所有的数据进行训练。相当于做了一次cross_val_score