K-fold cross-validation_validation

全栈程序员站长

发布于 2022-11-06 12:18:22

4120

发布于 2022-11-06 12:18:22

文章被收录于专栏：全栈程序员必看

大家好，又见面了，我是你们的朋友全栈君。

KFold是sklearn中用来做交叉检验的，在sklearn 的版本升级中，KFold被挪了地方。

在sklearn 0.18及以上的版本中，sklearn.cross_validation包被废弃，KFold被挪到了sklearn.model_selection中，本来以为挪就挪了，用法没变就行，结果，，谁用谁知道。

cross_validation.KFold与model_selection.KFold的不同用法

cross_validation.KFold做交叉验证

from sklearn.linear_model import LogisticRegression
from sklearn.cross_validation import KFold, cross_val_score
from sklearn.metrics import confusion_matrix,recall_score,classification_report

def printing_Kfold_scores(x_train_data,y_train_data):
    fold = KFold(len(y_train_data),5,shuffle=False) 
    #将训练集切分成5份，做交叉验证
  
    #正则化惩罚项系数
    c_param_range = [0.01,0.1,1,10,100]

    results_table = pd.DataFrame(index = range(len(c_param_range),2), columns = ['C_parameter','Mean recall score'])
    results_table['C_parameter'] = c_param_range

    j = 0
    for c_param in c_param_range:
        print('-------------------------------------------')
        print('C parameter: ', c_param)
        print('-------------------------------------------')
        print('')

        recall_accs = []
        #循环进行交叉验证
        for iteration, indices in enumerate(fold,start=1):
            #建立逻辑回归模型，选择正则惩罚类型L1
            lr = LogisticRegression(C = c_param, penalty = 'l1')

            lr.fit(x_train_data.iloc[indices[0],:],y_train_data.iloc[indices[0],:].values.ravel())

            y_pred_undersample = lr.predict(x_train_data.iloc[indices[1],:].values)

            recall_acc = recall_score(y_train_data.iloc[indices[1],:].values,y_pred_undersample)#计算召回率
            
            recall_accs.append(recall_acc)
            
            print('Iteration ', iteration,': recall score = ', recall_acc)

        results_table.ix[j,'Mean recall score'] = np.mean(recall_accs)
        j += 1
        print('')
        print('Mean recall score ', np.mean(recall_accs))
        print('')
    results_table['Mean recall score'] = results_table['Mean recall score'].astype('float64')
    best_c = results_table.loc[results_table['Mean recall score'].idxmax()]['C_parameter']
    
    # Finally, we can check which C parameter is the best amongst the chosen.
    print('*********************************************************************************')
    print('Best model to choose from cross validation is with C parameter = ', best_c)
    print('*********************************************************************************')
    
    return best_c

model_selection.KFold做交叉验证

from sklearn.linear_model import LogisticRegression
from sklearn.model_selection import KFold, cross_val_score
from sklearn.metrics import confusion_matrix,recall_score,classification_report 

def printing_Kfold_scores(x_train_data,y_train_data):
    #将训练集切分成5份，做交叉验证
    kf = KFold(n_splits=5,shuffle=False)
    kf.get_n_splits(x_train_data)

    #正则化惩罚项系数
    c_param_range = [0.01,0.1,1,10,100]

    results_table = pd.DataFrame(index = range(len(c_param_range),2), columns = ['C_parameter','Mean recall score'])
    results_table['C_parameter'] = c_param_range

    j = 0
    for c_param in c_param_range:
        print('-------------------------------------------')
        print('C parameter: ', c_param)
        print('-------------------------------------------')
        print('')

        recall_accs = []
        #循环进行交叉验证
        for iteration, indices in kf.split(x_train_data):
            
            lr = LogisticRegression(C = c_param, penalty = 'l1',solver='liblinear')
            
            lr.fit(x_train_data.iloc[iteration,:],y_train_data.iloc[iteration,:].values.ravel())

            y_pred_undersample = lr.predict(x_train_data.iloc[indices,:].values)

            recall_acc = recall_score(y_train_data.iloc[indices,:].values,y_pred_undersample)#计算召回率
            recall_accs.append(recall_acc)
            
            print('recall score = ', recall_acc)

        results_table.ix[j,'Mean recall score'] = np.mean(recall_accs)
        j += 1
        print('')
        print('Mean recall score ', np.mean(recall_accs))
        print('')
    results_table['Mean recall score'] = results_table['Mean recall score'].astype('float64')
    best_c = results_table.loc[results_table['Mean recall score'].idxmax()]['C_parameter']
    
    # Finally, we can check which C parameter is the best amongst the chosen.
    print('*********************************************************************************')
    print('Best model to choose from cross validation is with C parameter = ', best_c)
    print('*********************************************************************************')
    
    return best_c

在新版中，将数据切分需要两行代码：kf = KFold(n_splits=5,shuffle=False) 、 kf.get_n_splits(x_train_data)，用for iteration, indices in kf.split(x_train_data):取出，看到iteration和indices装的是两段index值，iteration装了五分之四，indices装的是五分之一，如下图

在旧版本中，将数据切分成n份就是一句代码：fold = KFold(len(y_train_data),5,shuffle=False)，并且切分后用：for iteration, indices in enumerate(fold,start=1):，取出的iteration是1、2、3、4、5这几个数，indices是上图中两部分的合集

本文参与腾讯云自媒体同步曝光计划，分享自作者个人站点/博客。

原始发表：2022年9月20日，如有侵权请联系 cloudcommunity@tencent.com 删除

spring