首先需要介紹的工具是sklearn的模型選擇API(GridSearchCV)

第一節：GridSearchCV函式的用法

sklearn.grid_search.GridSearchCV(
	estimator, # 是你要訓練的模型booster
	param_grid, # 字典型別的params 需要選擇的超參
	scoring=None, # 評判效能好壞的標準 如auc: scoring='roc_auc'
	fit_params=None, 
	n_jobs=1, # 並行執行的作業數
	iid=True, 
	refit=True, 
	cv=None, # 訓練集與驗證集的比值，相當於nfold=5
	verbose=0, 
	pre_dispatch='2*n_jobs', 
	error_score='raise'
)
 

#-*- coding:utf-8 -*-
import pandas as pd
from sklearn import svm, datasets
from sklearn.model_selection import GridSearchCV
from sklearn.metrics import classification_report

iris = datasets.load_iris()
# 待調整的超引數
parameters = {
	'kernel':('linear', 'rbf'), 
	'C':[1, 2, 4], 
	'gamma':[0.125, 0.25, 0.5 ,1, 2, 4]
}
svr = svm.SVC() # 模型
clf = GridSearchCV( svr, parameters, n_jobs=4 )
clf.fit(iris.data, iris.target) #你和模型
# clf.cv_results_ 是選擇引數的日誌資訊
cv_result = pd.DataFrame.from_dict( clf.cv_results_ )
with open('./data/cv_result.csv','wb') as f: cv_result.to_csv(f)
print 'The parameters of the best model are: '
print  clf.best_params_  # 打印出最合適的模型引數
y_pred_array = clf.predict(iris.data) # 預測結果
print classification_report( y_true=iris.target, y_pred=y_pred_array )
 

{'kernel': 'linear', 'C': 2, 'gamma': 0.125}

print classification_report( y_true=iris.target, y_pred=y_pred_array ) 會列印如下結果

             precision    recall  f1-score   support

          0       1.00      1.00      1.00        50
          1       1.00      0.94      0.97        50
          2       0.94      1.00      0.97        50

avg / total       0.98      0.98      0.98       150
 

第二節：lightGBM使用GridSearchCV調參

lightgbm.sklearn.LGBMRegressor( 
    boosting_type='gbdt', 
    num_leaves=31, 
    max_depth=-1, 
    learning_rate=0.1, 
    n_estimators=10, 
    max_bin=255, 
    subsample_for_bin=50000, 
    objective='regression', 
    min_split_gain=0, 
    min_child_weight=5, 
    min_child_samples=10, 
    subsample=1, 
    subsample_freq=1, 
    colsample_bytree=1, 
    reg_alpha=0, 
    reg_lambda=0, 
    seed=0, 
    nthread=-1, 
    silent=True, 
    huber_delta=1.0, 
    gaussian_eta=1.0, 
    fair_c=1.0, 
    poisson_max_delta_step=0.7, 
    drop_rate=0.1, 
    skip_drop=0.5, 
    max_drop=50, 
    uniform_drop=False, 
    xgboost_dart_mode=False
)