贝叶斯调参

    from bayes_opt import BayesianOptimization      # 定义黑盒函数     def lgb_cv(n_estimators,learning_rate,max_depth,num_leaves,min_child_samples,subsample,colsample_bytree,reg_alpha,reg_lambda,min_child_weight):         val = cross_val_score(             LGBMClassifier(n_estimators=int(n_estimators),                            random_state=2019,                            learning_rate=learning_rate,                            max_depth=int(max_depth),                            num_leaves=int(num_leaves),                            min_child_samples=int(min_child_samples),                            subsample=min(subsample, 0.99999),                            colsample_bytree=min(colsample_bytree, 0.9999),                            reg_alpha=reg_alpha,                            reg_lambda=reg_lambda,                            min_child_weight=min_child_weight,                            class_weight='balanced',                            ), train_x[use_col], train_y, scoring='roc_auc', cv=5,n_jobs=-1         ).mean()         return val      # 给定超参数搜索空间     opt = BayesianOptimization(         lgb_cv,         {'n_estimators': (600, 1000),          'learning_rate':(0.005,0.015),          'max_depth':(2,5),          'num_leaves':(6,16),          'min_child_samples':(10,50),          'subsample':(0.6,0.99),          'colsample_bytree':(0.6,0.99),          'reg_alpha':(0.05,1),          'reg_lambda':(0.05,1),          'min_child_weight':(1,100)          }     )     opt.maximize()  # 最大化黑盒函数       opt.max   # 返回黑盒函数值最大的超参数     

来源：博客园

作者：洋洋同学

链接：https://www.cnblogs.com/lky520hs/p/11770135.html