# 如果运行报错:ValueError: The number of classes has to be greater than one; got 1 class
# 直接多运行几次或者多等待一会

# 数据读取
import pandas as pd
import numpy as np  # 注意表头的位置
import json


import os  # 清除残余文件
import pickle
# 清楚文件
if os.path.exists('model/ecg_svm.pkl'):
    os.remove('model/ecg_svm.pkl')

if os.path.exists('model/ecg_knn.pkl'):
    os.remove('model/ecg_knn.pkl')

if os.path.exists('model/ecg_rand.pkl'):
    os.remove('model/ecg_rand.pkl')

if os.path.exists('model/ecg_adbost.pkl'):
    os.remove('model/ecg_adbost.pkl')


logname = 'train_report/log_ecg.txt'
with open(logname, 'w', encoding='utf-8', newline='') as f:
    print('********************运行日志**************************', file=f)

    csv = pd.read_csv('feature.csv', header=0)  # 筛选特征
    data_x = csv[["bpm",
                  "ibi",
                  "sdnn",
                  "sdsd",
                  "rmssd",
                  "pnn20",
                  "pnn50",
                  "hr_mad",
                  "sd1",
                  "sd2",
                  "s",
                  "sd1/sd2",
                  "breathingrate"
                  ]]
    data_y = np.array(csv[['label']])[:, 0]  # 降维
    # 注: label 里的值至少有两个不同的, 否则向量机报错

    print('***********************读取文件成功,分类开始*************************\n', file=f)

    print('数据集划分:\n', file=f)

    # 数据集划分
    from sklearn.model_selection import train_test_split
    x_train, x_test, y_train, y_test = train_test_split(data_x, data_y, test_size=0.1)  # 验证样本比例

    print('训练数据条目: %d  训练标签条目:%d  验证数据条目:%d  验证标签条目:%d\n' % (len(x_train), len(y_train), len(x_test), len(y_test)), file=f)

    from sklearn.preprocessing import StandardScaler   # 数据标准化
    transfer = StandardScaler()
    x_train = transfer.fit_transform(x_train)
    x_test = transfer.fit_transform(x_test)

    print('数据标准化完成!\n', file=f)

    from sklearn import svm   # 支持向量机
    clf = svm.SVC()
    clf.fit(x_train, y_train)
    # 保存网络模型
    with open('model/ecg_svm.pkl', 'wb') as fw:
        pickle.dump(clf, fw)

    print('ecg_svm.pkl 模型保存成功! \n', file=f)
    # 读取测试
    # with open('model/ecg_svm.pkl', 'rb') as fr:
    #  new_svm = pickle.load(fr)
    #  print (new_svm.predict(x_test))

    print('----------------对比预测结果-----------------', file=f)

    print('----------------对比预测结果-----------------', file=f)

    # 写入字典

    print(y_test == clf.predict(x_test), file=f)  # 直接对比预测结果
    print('-------------------------------------------', file=f)
    accuracy = clf.score(x_test, y_test) * 100  # 准确率计算

    print('准确率为:%f\n' % (accuracy), file=f)

    print('*********************支持向量机分类结束***********************\n', file=f)

    print('\n', file=f)

    print('***********************开始KNN分类**************************\n', file=f)

    from sklearn.neighbors import KNeighborsClassifier  # KNN算法       #分类蔟
    classifier = KNeighborsClassifier(n_neighbors=2, metric='minkowski', p=2)
    classifier.fit(x_train, y_train)
    # 保存网络模型
    with open('model/ecg_knn.pkl', 'wb') as fw:
        pickle.dump(classifier, fw)

    print('ecg_knn.pkl 模型保存成功! \n', file=f)
    y_pred = classifier.predict(x_test)  # Predicting the Test set results

    accuracy = classifier.score(x_test, y_test)

    print('---------------------生成KNN混肴矩阵-------------------\n', file=f)

    from sklearn.metrics import confusion_matrix   # Making the Confusion Matrix
    from sklearn.metrics import classification_report
    cm = confusion_matrix(y_test, y_pred)

    print(cm, file=f)
    print(classification_report(y_test, y_pred), file=f)
    print('---------------------生成完毕-----------------------\n', file=f)

    print('*********************KNN分类结束**********************************', file=f)

    print('\n', file=f)
    print('\n', file=f)

    print('***********************开始随机森林分类**************************\n', file=f)

    from sklearn.ensemble import RandomForestClassifier  # Fitting Random Forest to the Training set
    classifier = RandomForestClassifier(n_estimators=5, criterion='entropy', random_state=0)  # 随机选取点
    classifier.fit(x_train, y_train)
    # 保存网络模型
    with open('model/ecg_rand.pkl', 'wb') as fw:
        pickle.dump(classifier, fw)

    print('ecg_rand.pkl 模型保存成功! \n', file=f)
    y_pred = classifier.predict(x_test)  # Predicting the Test set results

    accuracy = classifier.score(x_test, y_test)

    print('预测验证数据集:', file=f)
    print(y_pred, file=f)
    print('\n', file=f)

    print('---------------------生成随机森林混肴矩阵-------------------\n', file=f)

    from sklearn.metrics import confusion_matrix  # Making the Confusion Matrix
    cm = confusion_matrix(y_test, y_pred)

    print(cm, file=f)
    print(classification_report(y_test, y_pred), file=f)
    print('---------------------生成完毕-----------------------\n', file=f)
    print('*********************随机森林分类结束**********************************', file=f)

    print('\n', file=f)
    print('\n', file=f)

    print('\n', file=f)
    print('\n', file=f)

    print('***********************开始AdaBoost分类**************************\n', file=f)
    from sklearn.ensemble import AdaBoostClassifier
    from sklearn import metrics
    # 定义模型                               #迭代次数         学习率
    model = AdaBoostClassifier(n_estimators=50, learning_rate=1.5)
    # 训练
    model.fit(x_train, y_train)
    # 保存网络模型
    with open('model/ecg_adbost.pkl', 'wb') as fw:
        pickle.dump(model, fw)

    print('ecg_adbost.pkl 模型保存成功! \n', file=f)
    # 预测
    y_pred = model.predict(x_test)

    print('预测验证数据集:', file=f)
    print(y_pred, file=f)
    print('\n', file=f)
    acc = metrics.accuracy_score(y_test, y_pred)  # 准确率

    print(f"准确率:{acc:.2}\n", file=f)
    print('***********************AdaBoost分类结束**************************\n', file=f)


print('心电数据训练结束\n')