Metadata-Version: 2.1
Name: stackingNT
Version: 0.0.1
Summary: Stacking algorithm optimisation
Home-page: UNKNOWN
Author: caoxinyu
Author-email: 398424191@qq.com
License: UNKNOWN
Keywords: python,menu,dumb_menu,windows,mac,linux
Platform: UNKNOWN
Classifier: Development Status :: 1 - Planning
Classifier: Intended Audience :: Developers
Classifier: Programming Language :: Python :: 3
Classifier: Operating System :: Unix
Classifier: Operating System :: MacOS :: MacOS X
Classifier: Operating System :: Microsoft :: Windows
Description-Content-Type: text/markdown


```python

# 数据导入

import numpy as np

import pandas as pd



train_path = r"C:\Users\Admin\Desktop\data\miccai_data\train_P_value_8.xlsx"

test_path = r"C:\Users\Admin\Desktop\data\miccai_data\test_P_value_8.xlsx"



train_df = pd.read_excel(train_path)

test_df = pd.read_excel(test_path)



train_df_lassoCV = train_df[train_df.columns[2:]]

Y = train_df['综合指南与治疗的分类方法']



test_df_lassoCV = test_df[test_df.columns[2:]]

Y_test = test_df['综合指南与治疗的分类方法']





```





```python

# 建立模型池



# 机器学习分类模型建模

from sklearn.svm import SVC

from sklearn.ensemble import RandomForestClassifier

from xgboost import XGBClassifier

from sklearn.tree import DecisionTreeClassifier

from sklearn.ensemble import AdaBoostClassifier

from sklearn.linear_model import LogisticRegression





random_seed = 2022

n_classes = [0, 1, 2]

classifiers_ = {

    'RandomForest': RandomForestClassifier(random_state=random_seed, n_estimators=100, max_depth=None),

    'DecisionTree': DecisionTreeClassifier(random_state=random_seed),

    'XGBoost': XGBClassifier(reg_lambda=0.5,

                             max_depth=8,

                             learning_rate=0.93,

                             n_estimators=100, 

                             min_child_weight=1,

                             gamma=0.3,

                             # min_weight=5,

                             colsample_bytree=0.8,

                             verbosity=0,

                             num_class=len(n_classes),

                             objective='multi:softmax',

                             random_state=random_seed),

    # 'AdaBoost': AdaBoostClassifier(n_estimators=100, learning_rate=0.9, random_state=random_seed),

    'LogisticRegression':LogisticRegression(random_state=random_seed),

    'SVM': SVC(kernel='linear', C=1, probability=True, tol=1.e-4, random_state=random_seed),

}



```



    d:\Anaconda\envs\Tensorflow\lib\site-packages\scipy\__init__.py:146: UserWarning: A NumPy version >=1.16.5 and <1.23.0 is required for this version of SciPy (detected version 1.24.2

      warnings.warn(f"A NumPy version >={np_minversion} and <{np_maxversion}"

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\compat.py:31: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead.

      from pandas import MultiIndex, Int64Index

    





```python

from sklearn.preprocessing import label_binarize

# 更新样本标签为多分类形式

Y_train_multi = label_binarize(Y, classes=n_classes)

Y_test_multi = label_binarize(Y_test, classes=n_classes)



```





```python

# 观察机器学习模型在训练集/测试集上的性能

for name in classifiers_:

    model = classifiers_[name]

    model.fit(train_df_lassoCV, Y)

    print("模型 {} 在训练集上的性能(ACC)为 {}".format(name, model.score(train_df_lassoCV, Y)))



    print("模型 {} 在测试集上的性能(ACC)为 {}".format(name, model.score(test_df_lassoCV, Y_test)))

print()

```



    模型 RandomForest 在训练集上的性能(ACC)为 1.0

    模型 RandomForest 在测试集上的性能(ACC)为 0.625

    模型 DecisionTree 在训练集上的性能(ACC)为 1.0

    模型 DecisionTree 在测试集上的性能(ACC)为 0.7

    模型 XGBoost 在训练集上的性能(ACC)为 1.0

    模型 XGBoost 在测试集上的性能(ACC)为 0.7

    模型 LogisticRegression 在训练集上的性能(ACC)为 0.6242038216560509

    模型 LogisticRegression 在测试集上的性能(ACC)为 0.6

    模型 SVM 在训练集上的性能(ACC)为 0.6305732484076433

    模型 SVM 在测试集上的性能(ACC)为 0.55

    

    



    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\data.py:208: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead.

      from pandas import MultiIndex, Int64Index

    





```python

# 训练集上10折交叉验证的性能

from sklearn.model_selection import cross_validate



best_models = dict()

for name in classifiers_:

    model = classifiers_[name]

    cv_results = cross_validate(model, train_df_lassoCV, Y.values, cv=10, return_estimator=True,

                                return_train_score=True)

    test_results = []

    for m in cv_results['estimator']:

        test_results.append(m.score(test_df_lassoCV, Y_test.values))



    # 保存在测试集上性能最好的模型

    ind = np.argmax(test_results)

    best_models[name] = cv_results['estimator'][ind]

    print("模型 {} 经过10折交叉验证的性能(ACC)为 {:.4f}（训练集）{:.4f}（验证集）{:.4f}（测试集）".format(name, cv_results["train_score"].mean(),

                                                                              cv_results["test_score"].mean(),

                                                                              test_results[ind]))





```



    模型 RandomForest 经过10折交叉验证的性能(ACC)为 1.0000（训练集）0.6617（验证集）0.6750（测试集）

    模型 DecisionTree 经过10折交叉验证的性能(ACC)为 1.0000（训练集）0.6875（验证集）0.7500（测试集）

    



    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\data.py:208: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead.

      from pandas import MultiIndex, Int64Index

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\data.py:208: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead.

      from pandas import MultiIndex, Int64Index

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\data.py:208: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead.

      from pandas import MultiIndex, Int64Index

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\data.py:208: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead.

      from pandas import MultiIndex, Int64Index

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\data.py:208: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead.

      from pandas import MultiIndex, Int64Index

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\data.py:208: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead.

      from pandas import MultiIndex, Int64Index

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\data.py:208: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead.

      from pandas import MultiIndex, Int64Index

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\data.py:208: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead.

      from pandas import MultiIndex, Int64Index

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\data.py:208: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead.

      from pandas import MultiIndex, Int64Index

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\data.py:208: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead.

      from pandas import MultiIndex, Int64Index

    



    模型 XGBoost 经过10折交叉验证的性能(ACC)为 1.0000（训练集）0.6421（验证集）0.6750（测试集）

    模型 LogisticRegression 经过10折交叉验证的性能(ACC)为 0.6376（训练集）0.6029（验证集）0.6250（测试集）

    模型 SVM 经过10折交叉验证的性能(ACC)为 0.6192（训练集）0.5842（验证集）0.6250（测试集）

    





```python

def metrics_multiclass(cm, average="macro"):

    """

    计算多分类任务的sensitivity和specificity

      average: {"macro", "micro"} 用于多分类的性能参数计算方式

        'micro':

        Calculate metrics globally by counting the total true positives, false negatives and false positives.

        'macro':

        Calculate metrics for each label, and find their unweighted mean. This does not take label imbalance into account.

    """

    n_classes = cm.shape[0]

    sen_tmp = []

    spe_tmp = []

    tp_tmp = 0

    fn_tmp = 0

    tn_tmp = 0

    fp_tmp = 0

    for i in range(n_classes):

        # 逐步获取 真阳，假阳，真阴，假阴四个指标，并计算三个参数

        ALL = np.sum(cm)

        # 对角线上是正确预测的

        TP = cm[i, i]

        # 列加和减去正确预测是该类的假阳

        FP = np.sum(cm[:, i]) - TP

        # 行加和减去正确预测是该类的假阴

        FN = np.sum(cm[i, :]) - TP

        # 全部减去前面三个就是真阴

        TN = ALL - TP - FP - FN



        # 累积计算

        tp_tmp = tp_tmp + TP

        fp_tmp = fp_tmp + FP

        fn_tmp = fn_tmp + FN

        tn_tmp = tn_tmp + TN



        sen_tmp.append(TP / (TP + FN))

        spe_tmp.append(TN / (TN + FP))



    if average == "macro":

        sen = np.average(sen_tmp)

        spe = np.average(spe_tmp)

    else:

        sen = tp_tmp / (tp_tmp + fn_tmp)

        spe = tn_tmp / (tn_tmp + fp_tmp)



    return sen, spe

```





```python

# 绘制训练集和测试集上的ROC曲线 （多分类计算OVR的Macro AUC）

#

#   参考 https://scikit-learn.org.cn/view/295.html

#

from sklearn.metrics import roc_curve, auc

from sklearn.metrics import confusion_matrix, f1_score

import pickle



# 绘制训练集和测试集上的ROC曲线 （多分类计算OVR的Macro AUC）

#

#   参考 https://scikit-learn.org.cn/view/295.html

#

from sklearn.metrics import roc_curve, auc

import matplotlib.pyplot as plt

from sklearn.metrics import confusion_matrix, f1_score

import pickle



colors = ['aqua', 'darkorange', 'cornflowerblue', 'deeppink', 'green']

plt.figure(figsize=(14, 8))

ax1 = plt.subplot(121)

ax2 = plt.subplot(122)

train_auc = []

test_auc = []

for i, name in enumerate(best_models):

    model = best_models[name]

    train_pred = model.predict(train_df_lassoCV)

    test_pred = model.predict(test_df_lassoCV)

    train_pred_prob = model.predict_proba(train_df_lassoCV)

    test_pred_prob = model.predict_proba(test_df_lassoCV)



    # 注意采用标签采用多分类的softmax形式

    train_fpr, train_tpr, _ = roc_curve(Y_train_multi.ravel(), train_pred_prob.ravel())

    train_auc.append(auc(train_fpr, train_tpr))

    ax1.plot(train_fpr, train_tpr, color=colors[i], linestyle='-', linewidth=2,

             label="{} (AUC={:.4f})".format(name, auc(train_fpr, train_tpr)))



    test_fpr, test_tpr, _ = roc_curve(Y_test_multi.ravel(), test_pred_prob.ravel())

    test_auc.append(auc(test_fpr, test_tpr))

    ax2.plot(test_fpr, test_tpr, color=colors[i], linestyle='-', linewidth=2,

             label="{} (AUC={:.4f})".format(name, auc(test_fpr, test_tpr)))



    # 计算f1 score, sensitivity和specificity

    train_f1 = f1_score(Y, train_pred, average='macro')

    test_f1 = f1_score(Y_test, test_pred, average='macro')



    train_cm = confusion_matrix(Y, train_pred)

    test_cm = confusion_matrix(Y_test, test_pred)

    train_sen, train_spe = metrics_multiclass(train_cm)

    test_sen, test_spe = metrics_multiclass(test_cm)



    print("模型 {} 经过10折交叉验证的性能为".format(name))

    print("        AUC / f1 / sensitivity / specificity")

    print("训练集 - {:.4f} {:.4f} {:.4f} {:.4f}".format(auc(train_fpr, train_tpr), train_f1, train_sen, train_spe))

    print("测试集 - {:.4f} {:.4f} {:.4f} {:.4f}\n".format(auc(test_fpr, test_tpr), test_f1, test_sen, test_spe))



ax1.plot([0, 1], [0, 1], color='gray', lw=2, linestyle='--')

ax2.plot([0, 1], [0, 1], color='gray', lw=2, linestyle='--')

ax1.set_xlim([0.0, 1.0])

ax1.set_ylim([0.0, 1.05])

ax2.set_xlim([0.0, 1.0])

ax2.set_ylim([0.0, 1.05])

ax1.legend(loc=4)

ax2.legend(loc=4)

plt.show()



```



    模型 RandomForest 经过10折交叉验证的性能为

            AUC / f1 / sensitivity / specificity

    训练集 - 0.9974 0.9480 0.9495 0.9788

    测试集 - 0.8037 0.6574 0.6667 0.8411

    

    模型 DecisionTree 经过10折交叉验证的性能为

            AUC / f1 / sensitivity / specificity

    训练集 - 0.9761 0.9638 0.9653 0.9846

    测试集 - 0.8125 0.7467 0.7611 0.8744

    

    模型 XGBoost 经过10折交叉验证的性能为

            AUC / f1 / sensitivity / specificity

    训练集 - 0.9940 0.9704 0.9691 0.9879

    测试集 - 0.7866 0.6242 0.6463 0.8411

    

    模型 LogisticRegression 经过10折交叉验证的性能为

            AUC / f1 / sensitivity / specificity

    训练集 - 0.8273 0.5979 0.5998 0.8120

    测试集 - 0.7719 0.6157 0.6204 0.8051

    

    模型 SVM 经过10折交叉验证的性能为

            AUC / f1 / sensitivity / specificity

    训练集 - 0.8132 0.5566 0.5607 0.7988

    测试集 - 0.7694 0.6196 0.6352 0.8189

    

    





    

![png](3_files/3_6_1.png)

    







```python

# 定义非线性变换接口

def sigmoid(x):

    return 1.0 / (1 + np.exp(-x))



def relu(x):

    return x



def tanh(x):

    return (np.exp(x) - np.exp(-x)) / (np.exp(x) + np.exp(-x))



def sin(x):

    return np.sin(x)



def cos(x):

    return np.cos(x)



def softmax(x):

    exps = np.exp(x - np.max(x, axis=-1, keepdims=True))

    return exps / np.sum(exps, axis=-1, keepdims=True)



def softplus(x):

    return np.log(1 + np.exp(x))



def elu(x, alpha=1.0):

    return np.where(x < 0, alpha * (np.exp(x) - 1), x)



def swish(x):

    return x * sigmoid(x)



def mish(x):

    return x * np.tanh(softplus(x))



def gelu(x):

    return 0.5 * x * (1 + np.tanh(np.sqrt(2 / np.pi) * (x + 0.044715 * np.power(x, 3))))



```





```python

# 定义Stacking+NT函数：

def Ensemble_add_feature(train, test, best_models):

    

    train_ = np.zeros((train.shape[0], len(best_models ) + len(train_df_lassoCV.columns) ))

    test_ = np.zeros((test.shape[0], len(best_models)+len(test_df_lassoCV.columns) ))

    print(len(train_df_lassoCV))

    train_[:,0:len(train_df_lassoCV.columns)] = train_df_lassoCV

    test_[:,0:len(test_df_lassoCV.columns)] = test_df_lassoCV



    for i, name in enumerate(best_models):

        model = best_models[name]

        train_pred = model.predict(train)

        test_pred = model.predict(test)

    # train_pred_prob = model.predict_proba(train_df_lassoCV)

    # test_pred_prob = model.predict_proba(test_df_lassoCV)   

        ## 新特征生成

        # train_[:, len(train_df_lassoCV) + i] = train_pred ** 2

        # test_[:, len(train_df_lassoCV) + i] = test_pred ** 2



        # train_[:, len(best_models ) +i] = np.exp(train_pred)

        # test_[:, len(best_models ) +i] = np.exp(test_pred)



        train_[:, len(train_df_lassoCV.columns)+i] = sigmoid(train_pred)

        test_[:,len(train_df_lassoCV.columns)+i] = sigmoid(test_pred)



        # train_[:, len(train_df_lassoCV.columns)+i] = sin(train_pred)

        # test_[:, len(train_df_lassoCV.columns)+i] = sin(test_pred)



        # train_[:, len(train_df_lassoCV.columns)+i] = cos(train_pred)

        # test_[:, len(train_df_lassoCV.columns)+i] = cos(test_pred)



        # train_[:,len(train_df_lassoCV.columns)+i] = tanh(train_pred)

        # test_[:,len(train_df_lassoCV.columns)+i] = tanh(test_pred)



        # train_[:, len(train_df_lassoCV.columns)+i] = relu(train_pred)

        # test_[:, len(train_df_lassoCV.columns)+i] = relu(test_pred)



        # softmax 效果不好

        # train_[:, len(train_df_lassoCV.columns)+i] = softmax(train_pred)

        # test_[:, len(train_df_lassoCV.columns)+i] = softmax(test_pred)



        # softplus 

        # train_[:, len(train_df_lassoCV.columns)+i] = softplus(train_pred)

        # test_[:, len(train_df_lassoCV.columns)+i] = softplus(test_pred)





        # train_[:, len(train_df_lassoCV.columns)+i] = elu(train_pred)

        # test_[:, len(train_df_lassoCV.columns)+i] = elu(test_pred)



        # train_[:, len(train_df_lassoCV.columns)+i] = swish(train_pred)

        # test_[:, len(train_df_lassoCV.columns)+i] = swish(test_pred)



        # train_[:, len(train_df_lassoCV.columns)+i] = mish(train_pred)

        # test_[:, len(train_df_lassoCV.columns)+i] = mish(test_pred)



        # train_[:, len(train_df_lassoCV.columns)+i] = gelu(train_pred)

        # test_[:, len(train_df_lassoCV.columns)+i] = gelu(test_pred)



    

    return train_, test_

```





```python

new_train, new_test =  Ensemble_add_feature(train_df_lassoCV, test_df_lassoCV, best_models)

```



    157

    





```python

# 训练集上10折交叉验证的性能

from sklearn.model_selection import cross_validate



new_best_models = dict()

cv = 10

for name in classifiers_:

    model = classifiers_[name]

    cv_results = cross_validate(model, new_train, Y.values, cv=cv, return_estimator=True,

                                return_train_score=True)

    test_results = []

    for m in cv_results['estimator']:

        test_results.append(m.score(new_test, Y_test.values))



    # 保存在测试集上性能最好的模型

    ind = np.argmax(test_results)

    new_best_models[name] = cv_results['estimator'][ind]

    print("模型 {} 经过10折交叉验证的性能(ACC)为 {:.4f}（训练集）{:.4f}（验证集）{:.4f}（测试集）".format(name, cv_results["train_score"].mean(),

                                                                              cv_results["test_score"].mean(),

                                                                              test_results[ind]))

print()



new_train_auc = []

new_test_auc = []

for i, name in enumerate(new_best_models):

    model = new_best_models[name]

    train_pred = model.predict(new_train)

    test_pred = model.predict(new_test)

    train_pred_prob = model.predict_proba(new_train)

    test_pred_prob = model.predict_proba(new_test)



    # ACC

    print("模型 {} 在训练集上的性能(ACC)为 {}".format(name, model.score(new_train, Y)))

    print("模型 {} 在测试集上的性能(ACC)为 {}".format(name, model.score(new_test, Y_test)))



    # 注意采用标签采用多分类的softmax形式

    train_fpr, train_tpr, _ = roc_curve(Y_train_multi.ravel(), train_pred_prob.ravel())

    new_train_auc.append(auc(train_fpr, train_tpr))

    # ax1.plot(train_fpr, train_tpr, color=colors[i], linestyle='-', linewidth=2,

    #          label="{} (AUC={:.4f})".format(name, auc(train_fpr, train_tpr)))



    test_fpr, test_tpr, _ = roc_curve(Y_test_multi.ravel(), test_pred_prob.ravel())

    new_test_auc.append(auc(test_fpr, test_tpr))

    # ax2.plot(test_fpr, test_tpr, color=colors[i], linestyle='-', linewidth=2,

    #          label="{} (AUC={:.4f})".format(name, auc(test_fpr, test_tpr)))



    # 计算f1 score, sensitivity和specificity

    train_f1 = f1_score(Y, train_pred, average='macro')

    test_f1 = f1_score(Y_test, test_pred, average='macro')



    train_cm = confusion_matrix(Y, train_pred)

    test_cm = confusion_matrix(Y_test, test_pred)

    train_sen, train_spe = metrics_multiclass(train_cm)

    test_sen, test_spe = metrics_multiclass(test_cm)



    print("模型 {} 经过10折交叉验证的性能为".format(name))

    print("        AUC / f1 / sensitivity / specificity")

    print("训练集 - {:.4f} {:.4f} {:.4f} {:.4f}".format(auc(train_fpr, train_tpr), train_f1, train_sen, train_spe))

    print("测试集 - {:.4f} {:.4f} {:.4f} {:.4f}\n".format(auc(test_fpr, test_tpr), test_f1, test_sen, test_spe))

```



    模型 RandomForest 经过10折交叉验证的性能(ACC)为 1.0000（训练集）0.9433（验证集）0.7250（测试集）

    模型 DecisionTree 经过10折交叉验证的性能(ACC)为 1.0000（训练集）0.8979（验证集）0.7500（测试集）

    



    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    



    模型 XGBoost 经过10折交叉验证的性能(ACC)为 1.0000（训练集）0.8979（验证集）0.7500（测试集）

    模型 LogisticRegression 经过10折交叉验证的性能(ACC)为 0.9186（训练集）0.8904（验证集）0.7750（测试集）

    模型 SVM 经过10折交叉验证的性能(ACC)为 0.9816（训练集）0.9742（验证集）0.7750（测试集）

    

    模型 RandomForest 在训练集上的性能(ACC)为 1.0

    模型 RandomForest 在测试集上的性能(ACC)为 0.725

    模型 RandomForest 经过10折交叉验证的性能为

            AUC / f1 / sensitivity / specificity

    训练集 - 1.0000 1.0000 1.0000 1.0000

    测试集 - 0.8409 0.7020 0.7130 0.8617

    

    模型 DecisionTree 在训练集上的性能(ACC)为 0.9681528662420382

    模型 DecisionTree 在测试集上的性能(ACC)为 0.75

    模型 DecisionTree 经过10折交叉验证的性能为

            AUC / f1 / sensitivity / specificity

    训练集 - 0.9761 0.9638 0.9653 0.9846

    测试集 - 0.8125 0.7467 0.7611 0.8744

    

    模型 XGBoost 在训练集上的性能(ACC)为 0.9681528662420382

    模型 XGBoost 在测试集上的性能(ACC)为 0.75

    模型 XGBoost 经过10折交叉验证的性能为

            AUC / f1 / sensitivity / specificity

    训练集 - 0.9995 0.9638 0.9653 0.9846

    测试集 - 0.8644 0.7467 0.7611 0.8744

    

    模型 LogisticRegression 在训练集上的性能(ACC)为 0.9171974522292994

    模型 LogisticRegression 在测试集上的性能(ACC)为 0.775

    模型 LogisticRegression 经过10折交叉验证的性能为

            AUC / f1 / sensitivity / specificity

    训练集 - 0.9894 0.8986 0.8876 0.9537

    测试集 - 0.8562 0.7676 0.7796 0.8887

    

    模型 SVM 在训练集上的性能(ACC)为 0.9808917197452229

    模型 SVM 在测试集上的性能(ACC)为 0.775

    模型 SVM 经过10折交叉验证的性能为

            AUC / f1 / sensitivity / specificity

    训练集 - 0.9990 0.9800 0.9820 0.9905

    测试集 - 0.8703 0.7661 0.7796 0.8920

    

    





```python

# 训练集上10折交叉验证的性能

from sklearn.model_selection import cross_validate



best_models = dict()

for name in classifiers_:

    model = classifiers_[name]

    cv_results = cross_validate(model, train_df_lassoCV, Y.values, cv=10, return_estimator=True,

                                return_train_score=True)

    test_results = []

    for m in cv_results['estimator']:

        test_results.append(m.score(test_df_lassoCV, Y_test.values))



    # 保存在测试集上性能最好的模型

    ind = np.argmax(test_results)

    best_models[name] = cv_results['estimator'][ind]

    print("模型 {} 经过10折交叉验证的性能(ACC)为 {:.4f}（训练集）{:.4f}（验证集）{:.4f}（测试集）".format(name, cv_results["train_score"].mean(),

                                                                              cv_results["test_score"].mean(),

                                                                              test_results[ind]))



```





```python

# 定义一个class：

# 存储最优基模型的def

# 非线性变换的def

# 元模型进行训练的def

base_models = {}

meta_model = {}

from sklearn.model_selection import cross_validate



class StackingNonlinearTransformations():

    def __init__(self, base_models, meta_model, n_folds=10):

        self.base_models = base_models

        self.meta_model = meta_model

        self.n_folds = n_folds

        

   

    def NonlinearTransformations(self, x, NT, **kwargs):

        if NT == None:

            return x

        elif NT == "relu":

            return x

        elif NT == "sigmoid":

            return 1.0 / (1 + np.exp(-x))

        elif NT == "elu":

            if 'alpha' not in kwargs:

                raise ValueError('alpha is not given')

            alpha = kwargs["alpha"]

            return np.where(x < 0, alpha * (np.exp(x) - 1), x)

        elif NT == "tanh":

            return (np.exp(x) - np.exp(-x)) / (np.exp(x) + np.exp(-x))

        elif NT == "sin":

            return np.sin(x)

        elif NT == "cos":

            return np.cos(x)

        elif NT == "softmax":

            exps = np.exp(x - np.max(x, axis=-1, keepdims=True))

            return exps/np.sum(exps, axis=-1, keepdims=True)

        elif NT == "softplus":

            return np.log(1 + np.exp(x))





    def base_model_fit(self, train_X, train_y, test_X, test_y,cv=10):

        best_models = dict()

        for name in self.base_models.keys():

            model = self.base_models[name]

            cv_results = cross_validate(model, train_X, train_y.values, cv=cv, return_estimator=True, return_train_score=True)

            test_results = []

            for m in cv_results['estimator']:

                test_results.append(m.score(test_X, test_y.values))



            # 保存在测试集上性能最好的模型

            ind = np.argmax(test_results)

            best_models[name] = cv_results['estimator'][ind]

            







        train_ = np.zeros((train_X.shape[0], len(self.base_models ) + len(train_X.columns)))

        test_ = np.zeros((test_X.shape[0], len(self.base_models) + len(test_X.columns)))

        

        train_[:,0:len(train_X.columns)] = train_X

        test_[:,0:len(test_X.columns)] = test_X



        for i, name in enumerate(best_models):

            model = best_models[name]

            train_pred = model.predict(train_X)

            test_pred = model.predict(test_X)



            train_[:, len(train_X.columns)+i] = self.NonlinearTransformations(train_pred,"sigmoid")

            test_[:,len(test_X.columns)+i] = self.NonlinearTransformations(test_pred,"sigmoid")



            

        return train_, test_



    def fit(self, train_X, train_y, test_X, test_y, cv=10):



        new_train, new_test = self.base_model_fit(train_X, train_y, test_X, test_y,cv)

        model = list(self.meta_model.values())[0]

        cv_results = cross_validate(model, new_train, train_y.values, cv=cv, return_estimator=True, return_train_score=True)

        test_results = []

        for m in cv_results['estimator']:

            test_results.append(m.score(new_test, test_y.values))

        ind = np.argmax(test_results)

        new_best_models = cv_results['estimator'][ind]

        model = new_best_models

        train_pred = model.predict(new_train)

        test_pred = model.predict(new_test)

        train_pred_prob = model.predict_proba(new_train)

        test_pred_prob = model.predict_proba(new_test)

        

        return train_pred, test_pred, train_pred_prob, test_pred_prob







```





```python



meta_model = {'LogisticRegression':LogisticRegression(random_state=random_seed)}

base_models = classifiers_

SNT = StackingNonlinearTransformations(base_models, meta_model)

train_pred, test_pred, train_pred_prob, test_pred_prob = SNT.fit(train_X=train_df_lassoCV, train_y=Y, test_X=test_df_lassoCV, test_y=Y_test)



print(train_pred, test_pred, train_pred_prob, test_pred_prob)



from sklearn.metrics import roc_curve, auc

test_fpr, test_tpr, _ = roc_curve(Y_test_multi.ravel(), test_pred_prob.ravel())

print(auc(test_fpr, test_tpr))



```



    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\data.py:208: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead.

      from pandas import MultiIndex, Int64Index

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\data.py:208: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead.

      from pandas import MultiIndex, Int64Index

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\data.py:208: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead.

      from pandas import MultiIndex, Int64Index

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\data.py:208: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead.

      from pandas import MultiIndex, Int64Index

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\data.py:208: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead.

      from pandas import MultiIndex, Int64Index

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\data.py:208: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead.

      from pandas import MultiIndex, Int64Index

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\data.py:208: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead.

      from pandas import MultiIndex, Int64Index

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\data.py:208: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead.

      from pandas import MultiIndex, Int64Index

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\data.py:208: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead.

      from pandas import MultiIndex, Int64Index

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\sklearn.py:1146: UserWarning: The use of label encoder in XGBClassifier is deprecated and will be removed in a future release. To remove this warning, do the following: 1) Pass option use_label_encoder=False when constructing XGBClassifier object; and 2) Encode your labels (y) as integers starting with 0, i.e. 0, 1, 2, ..., [num_class - 1].

      warnings.warn(label_encoder_deprecation_msg, UserWarning)

    d:\Anaconda\envs\Tensorflow\lib\site-packages\xgboost\data.py:208: FutureWarning: pandas.Int64Index is deprecated and will be removed from pandas in a future version. Use pandas.Index with the appropriate dtype instead.

      from pandas import MultiIndex, Int64Index

    



    [0 0 0 0 0 2 2 2 2 2 2 2 2 1 2 2 2 2 2 2 0 0 0 1 0 0 1 0 2 2 2 1 0 1 2 0 2

     2 1 2 2 2 1 1 0 2 1 2 2 0 2 0 0 2 2 0 0 2 2 1 2 0 2 2 1 0 0 0 0 1 1 2 2 0

     0 1 2 2 2 0 2 2 2 0 1 0 0 0 0 2 2 1 0 2 0 2 1 2 0 2 2 1 1 0 2 2 2 2 2 0 2

     0 2 1 2 2 0 2 0 0 2 2 2 0 2 2 1 0 2 2 2 2 2 0 1 2 1 0 2 2 2 1 2 1 1 0 2 2

     0 2 2 0 0 2 1 2 0] [0 1 1 1 0 2 1 0 2 0 0 2 0 0 2 0 0 0 2 2 2 2 1 2 1 0 1 0 0 1 2 2 0 2 0 1 2

     2 1 2] [[7.98786397e-01 1.69605435e-01 3.16081676e-02]

     [8.03607989e-01 9.40798583e-02 1.02312152e-01]

     [7.52347690e-01 2.11163387e-01 3.64889232e-02]

     [7.91619966e-01 1.71735626e-01 3.66444083e-02]

     [8.66192574e-01 9.05399411e-02 4.32674849e-02]

     [2.64289472e-02 6.23805865e-02 9.11190466e-01]

     [6.58321922e-03 1.03126884e-02 9.83104092e-01]

     [5.56714002e-06 3.13986117e-07 9.99994119e-01]

     [6.31876006e-02 2.74163329e-01 6.62649071e-01]

     [9.03391815e-02 3.84241497e-01 5.25419322e-01]

     [1.29275193e-02 5.95725018e-02 9.27499979e-01]

     [2.73085475e-02 1.45869371e-01 8.26822082e-01]

     [4.35942768e-02 4.12456497e-01 5.43949227e-01]

     [1.85891093e-01 4.95132076e-01 3.18976831e-01]

     [2.21323474e-02 3.38608300e-02 9.44006823e-01]

     [1.87873110e-01 3.84692407e-01 4.27434483e-01]

     [2.68811165e-01 2.84499491e-01 4.46689344e-01]

     [2.59144723e-02 2.87414209e-01 6.86671319e-01]

     [2.06386607e-01 3.95290587e-01 3.98322806e-01]

     [3.62890421e-02 6.62503221e-02 8.97460636e-01]

     [7.87740453e-01 1.80174240e-01 3.20853068e-02]

     [7.00242210e-01 2.34947665e-01 6.48101245e-02]

     [7.37339484e-01 1.69966319e-01 9.26941969e-02]

     [2.77288314e-01 3.96972650e-01 3.25739036e-01]

     [6.69567838e-01 2.66316340e-01 6.41158219e-02]

     [7.47477430e-01 1.50726402e-01 1.01796167e-01]

     [8.70045602e-02 5.01211186e-01 4.11784254e-01]

     [8.68921831e-01 8.91031275e-02 4.19750416e-02]

     [3.25901714e-02 2.42757069e-01 7.24652760e-01]

     [3.53397592e-02 2.58418074e-01 7.06242167e-01]

     [6.81983930e-02 1.91605088e-01 7.40196518e-01]

     [2.29303699e-01 5.14796613e-01 2.55899688e-01]

     [7.28126469e-01 2.42286576e-01 2.95869548e-02]

     [2.11518519e-01 4.72495561e-01 3.15985920e-01]

     [6.86771750e-02 2.41202412e-01 6.90120413e-01]

     [4.56439625e-01 2.92334204e-01 2.51226171e-01]

     [2.35951003e-03 8.94287172e-03 9.88697618e-01]

     [5.84714019e-03 3.87784689e-02 9.55374391e-01]

     [2.32468741e-01 5.43872453e-01 2.23658805e-01]

     [1.01607216e-04 8.01645889e-05 9.99818228e-01]

     [7.37470020e-03 2.01742680e-02 9.72451032e-01]

     [1.75437722e-02 7.70095335e-02 9.05446694e-01]

     [1.51794269e-01 6.36902364e-01 2.11303367e-01]

     [2.28630779e-01 4.54513600e-01 3.16855621e-01]

     [7.45641217e-01 1.85859169e-01 6.84996133e-02]

     [6.52263123e-02 3.41753034e-01 5.93020654e-01]

     [2.01106012e-01 5.04824073e-01 2.94069915e-01]

     [6.56315924e-04 9.84681766e-04 9.98359002e-01]

     [4.66400475e-02 3.18332409e-01 6.35027543e-01]

     [8.23097715e-01 1.49822781e-01 2.70795036e-02]

     [2.41541421e-05 9.22716900e-06 9.99966619e-01]

     [4.46125562e-01 3.96839242e-01 1.57035197e-01]

     [7.68979315e-01 1.90777911e-01 4.02427742e-02]

     [1.67097338e-02 2.46627687e-02 9.58627497e-01]

     [6.55957679e-04 1.33265055e-03 9.98011392e-01]

     [7.89041958e-01 1.81510450e-01 2.94475925e-02]

     [4.76881159e-01 3.82695168e-01 1.40423673e-01]

     [1.38957721e-01 3.05852153e-01 5.55190126e-01]

     [5.78196942e-04 3.49984972e-05 9.99386805e-01]

     [3.68203364e-01 3.83338959e-01 2.48457677e-01]

     [6.08174150e-02 2.64501563e-01 6.74681022e-01]

     [9.11078059e-01 5.92273776e-02 2.96945630e-02]

     [3.69622029e-02 1.64157386e-01 7.98880411e-01]

     [7.58531105e-02 3.96294748e-01 5.27852141e-01]

     [1.95329341e-01 5.55302710e-01 2.49367949e-01]

     [8.00659609e-01 1.60786124e-01 3.85542674e-02]

     [6.98110394e-01 2.34441029e-01 6.74485774e-02]

     [7.65690794e-01 2.03125216e-01 3.11839894e-02]

     [7.16789649e-01 2.03575981e-01 7.96343699e-02]

     [2.52147856e-01 4.95772281e-01 2.52079863e-01]

     [2.33751947e-01 5.26615370e-01 2.39632682e-01]

     [3.48549836e-02 1.40844296e-01 8.24300720e-01]

     [4.31851108e-04 5.44385512e-05 9.99513710e-01]

     [7.08319324e-01 2.26379874e-01 6.53008018e-02]

     [7.85867768e-01 1.15875909e-01 9.82563230e-02]

     [1.49285446e-01 6.43352977e-01 2.07361577e-01]

     [4.77859030e-02 3.61083347e-01 5.91130750e-01]

     [1.32601339e-05 1.92751376e-05 9.99967465e-01]

     [8.69515058e-05 3.01424280e-05 9.99882906e-01]

     [7.73788515e-01 1.99475846e-01 2.67356393e-02]

     [2.52403851e-02 1.22695333e-01 8.52064282e-01]

     [6.62111189e-02 3.13246793e-01 6.20542088e-01]

     [3.99541391e-02 2.95868376e-01 6.64177485e-01]

     [8.22532255e-01 1.45832130e-01 3.16356146e-02]

     [2.63126961e-01 4.46488003e-01 2.90385035e-01]

     [8.58027770e-01 1.09849800e-01 3.21224294e-02]

     [7.10636412e-01 2.23726446e-01 6.56371416e-02]

     [8.12740585e-01 8.62682669e-02 1.00991148e-01]

     [8.19398349e-01 1.57370876e-01 2.32307747e-02]

     [5.96714067e-02 2.25097015e-01 7.15231578e-01]

     [2.92338228e-02 1.06116494e-01 8.64649683e-01]

     [1.11585307e-01 6.75251710e-01 2.13162983e-01]

     [8.31464270e-01 1.46748930e-01 2.17868003e-02]

     [6.74214901e-02 3.31668695e-01 6.00909815e-01]

     [7.00535288e-01 2.56306921e-01 4.31577905e-02]

     [1.42692784e-02 3.34527111e-03 9.82385450e-01]

     [1.77811887e-01 6.01371414e-01 2.20816699e-01]

     [4.86696249e-02 3.10275524e-01 6.41054851e-01]

     [7.75679989e-01 1.93388835e-01 3.09311763e-02]

     [2.16490047e-03 5.98298366e-05 9.97775270e-01]

     [1.24324901e-05 2.89506180e-06 9.99984672e-01]

     [2.22369795e-01 4.57757617e-01 3.19872588e-01]

     [2.05810905e-01 4.87066088e-01 3.07123008e-01]

     [8.74034657e-01 9.35941514e-02 3.23711917e-02]

     [3.67874915e-02 2.13302533e-01 7.49909975e-01]

     [3.59052095e-02 1.78350455e-01 7.85744335e-01]

     [2.20894555e-01 3.55160584e-01 4.23944861e-01]

     [5.24633391e-03 4.96205219e-02 9.45133144e-01]

     [5.27203325e-02 9.73688814e-02 8.49910786e-01]

     [8.17122662e-01 1.39414693e-01 4.34626442e-02]

     [3.42552164e-02 2.30473298e-01 7.35271485e-01]

     [8.01768720e-01 1.70115614e-01 2.81156656e-02]

     [2.91515710e-01 2.86419334e-01 4.22064956e-01]

     [3.01857693e-01 4.18977649e-01 2.79164658e-01]

     [1.31739300e-02 2.16117025e-02 9.65214368e-01]

     [5.00401636e-03 3.87830907e-03 9.91117675e-01]

     [7.44886182e-01 2.11559519e-01 4.35542991e-02]

     [5.95253556e-02 3.91207561e-01 5.49267083e-01]

     [8.66079551e-01 1.06102489e-01 2.78179601e-02]

     [3.65878132e-01 3.46774779e-01 2.87347090e-01]

     [7.18653042e-06 2.43778121e-06 9.99990376e-01]

     [8.86193319e-04 6.37648441e-05 9.99050042e-01]

     [2.39809501e-01 3.63195508e-01 3.96994992e-01]

     [7.48960624e-01 2.21827781e-01 2.92115956e-02]

     [3.70626424e-02 1.74146878e-01 7.88790480e-01]

     [3.35477348e-02 4.78102597e-01 4.88349669e-01]

     [1.81724731e-01 5.78662138e-01 2.39613131e-01]

     [8.44779549e-01 1.34731231e-01 2.04892198e-02]

     [4.29899032e-02 3.46649330e-01 6.10360767e-01]

     [1.25110049e-02 6.27672524e-02 9.24721743e-01]

     [2.54349517e-02 2.41452149e-01 7.33112900e-01]

     [4.66512604e-02 1.97634867e-01 7.55713873e-01]

     [2.65439747e-02 8.61988767e-02 8.87257149e-01]

     [7.16787257e-01 2.45763041e-01 3.74497015e-02]

     [1.95515717e-01 4.99840326e-01 3.04643957e-01]

     [2.31213016e-03 5.68899923e-03 9.91998871e-01]

     [3.02484695e-01 3.67696335e-01 3.29818969e-01]

     [6.03254895e-01 3.27745389e-01 6.89997166e-02]

     [4.84823524e-02 3.72184801e-01 5.79332847e-01]

     [3.88687483e-02 2.27865214e-01 7.33266038e-01]

     [3.72358130e-02 1.43826694e-01 8.18937493e-01]

     [4.22150974e-01 5.09967334e-01 6.78816923e-02]

     [1.44358511e-05 3.53400311e-06 9.99982030e-01]

     [2.93840559e-01 4.35732341e-01 2.70427100e-01]

     [3.08374499e-01 5.01335872e-01 1.90289628e-01]

     [5.50171673e-01 3.55933713e-01 9.38946137e-02]

     [1.00870996e-01 4.14954816e-01 4.84174188e-01]

     [2.43878907e-01 2.47910899e-01 5.08210194e-01]

     [7.84796219e-01 1.78989131e-01 3.62146502e-02]

     [6.17529014e-02 3.38641137e-01 5.99605961e-01]

     [2.32779319e-01 2.78395670e-01 4.88825011e-01]

     [6.23691713e-01 3.29952558e-01 4.63557298e-02]

     [6.48095343e-01 3.13361734e-01 3.85429224e-02]

     [2.98490899e-01 3.40569438e-01 3.60939664e-01]

     [2.47638702e-01 4.96969705e-01 2.55391593e-01]

     [3.59801681e-05 3.89351530e-05 9.99925085e-01]

     [8.22010958e-01 1.42218397e-01 3.57706458e-02]] [[7.57690319e-01 2.15765702e-01 2.65439787e-02]

     [2.24236188e-01 4.56931996e-01 3.18831817e-01]

     [4.51783852e-01 4.86295381e-01 6.19207666e-02]

     [2.67759413e-01 4.95781572e-01 2.36459015e-01]

     [8.93593736e-01 6.83144804e-02 3.80917834e-02]

     [4.56789989e-02 1.51049024e-01 8.03271977e-01]

     [3.48646077e-01 4.69162495e-01 1.82191428e-01]

     [7.70371069e-01 1.92453665e-01 3.71752660e-02]

     [2.92361720e-02 1.36400919e-01 8.34362909e-01]

     [7.91777546e-01 1.78397308e-01 2.98251461e-02]

     [7.05172587e-01 2.50739922e-01 4.40874910e-02]

     [6.44243061e-04 1.15470282e-03 9.98201054e-01]

     [8.33349584e-01 1.39406620e-01 2.72437955e-02]

     [7.94729027e-01 1.76119383e-01 2.91515906e-02]

     [7.62098585e-02 1.56504593e-01 7.67285548e-01]

     [6.26567093e-01 3.42803185e-01 3.06297217e-02]

     [7.47227843e-01 2.14844743e-01 3.79274140e-02]

     [7.61050009e-01 2.13900447e-01 2.50495442e-02]

     [3.24771495e-02 4.23211195e-01 5.44311656e-01]

     [1.18963573e-02 5.23715238e-02 9.35732119e-01]

     [2.13817593e-05 1.02287198e-05 9.99968390e-01]

     [4.50874319e-01 3.09577781e-02 5.18167903e-01]

     [2.77648676e-01 4.51336036e-01 2.71015288e-01]

     [1.28333994e-02 4.41898528e-02 9.42976748e-01]

     [1.91456101e-01 4.05978728e-01 4.02565170e-01]

     [6.58309566e-01 1.98781718e-01 1.42908716e-01]

     [2.24829634e-01 4.40746040e-01 3.34424326e-01]

     [6.29917199e-01 2.35113703e-01 1.34969098e-01]

     [7.66974837e-01 2.08536414e-01 2.44887488e-02]

     [4.16676999e-01 4.80316138e-01 1.03006862e-01]

     [1.58223891e-03 2.31608519e-03 9.96101676e-01]

     [2.58775164e-04 2.42472774e-05 9.99716978e-01]

     [8.40128641e-01 1.26973885e-01 3.28974738e-02]

     [7.21285970e-03 2.86309548e-02 9.64156185e-01]

     [6.52866666e-01 3.24764698e-01 2.23686356e-02]

     [2.41687444e-01 4.72064771e-01 2.86247785e-01]

     [8.28419375e-02 4.11308528e-01 5.05849534e-01]

     [4.40553422e-05 2.35332311e-05 9.99932411e-01]

     [2.78964612e-01 4.59652130e-01 2.61383258e-01]

     [2.97970311e-02 2.75009795e-02 9.42701989e-01]]

    0.85625

    

# upload log



0.0.4 更新正确的Class调用方法

