机器学习——数据处理流程梳理

数据处理流程梳理

  • 一、数据预处理
    • 1.数据缺失处理
    • 2.划分训练集、测试集及数据不平衡处理
    • 3.数据标准化
    • 4.特征选择
      • 1)随机森林
      • 2)相关性检测
      • 3)lasso特征
    • 5.特征降维
  • 二、训练模型
    • 1.训练
    • 2.查看ROC曲线
    • 3.评价指标
  • 三、测试数据预处理
    • 1.去除训练集缺失率高的列
    • 2.标准化
    • 3.特征选择
    • 4.特征降维
  • 四、测试集预测
  • 五、完整代码
  • 六、问题归纳
    • 1.数据集划分、数据平衡处理的顺序
    • 2.数据标准化应该放在哪一步
    • 3.关键参数
      • 1)行列缺失率row_miss_rate、col_miss_rate
      • 2)模型参数
      • 3)PCA保留的维度数
    • 4.roc曲线随测试集样本比例而变化,各种指标表现过低

一、数据预处理

以二分类问题为例,数据集形式为excle,label在最后一列

1.数据缺失处理

分别对行列缺失进行处理

def preTrainDataAndSelectFeature(self, file_path='', label_name='', col_miss_rate=0.5,row_miss_rate=0.8,
                              balance_prior=False,all_data_balance=0.33,train_data_balance=0.33,test_data_balance=None,
                              forest_n=None, kbest_n=None, lasso=False, pca_feature=4, ):
        """
        训练集数据处理
        :param file_path:string 文件路径
        :param label_name:标签列名
        :param col_miss_rate:float 列缺失比率
        :param row_miss_rate:float 标签为0的行缺失率
        :param balance_prior:bool 数据集平衡优先
        :param all_data_balance:balance_prior=True时,float 整体数据不均衡调整比例 若=1,为0,1标签比值为1:1,=0.33为1:0.33
        :param train_data_balance:balance_prior=False时,
        :param test_data_balance:balance_prior=False时,
        :param forest_n:int 使用forest选取的特征数
        :param kbest_n:int 使用kbest选取的特征数
        :param lasso:bool 是否使用lasso回归选取特征
        :param pca_feature:int pca降维后的特征数(这是降维的最后一步,数据此后一直使用降维后的特征)
        :return:可直接放入模型训练的最终数据
        """
        data = pd.read_excel(file_path)

        # 列处理
        data = data.dropna(axis=1, thresh=data.shape[0] * col_miss_rate)
        # 保存过滤缺失后的特征
        self.drop_select = list(data)
        self.final_feature = np.array(self.drop_select)
        print("drop select:",len(self.final_feature),"\n", self.final_feature)

        # 行处理:根据标签取值拆分样本,不同取值用各自的均值填充
        data_group = data.groupby(label_name)
        data0 = dict([x for x in data_group])[0]
        data0 = data0.dropna(axis=0, thresh=data0.shape[1] * row_miss_rate)
        data0 = data0.fillna(data0.mean())
        data1 = dict([x for x in data_group])[1]
        data1 = data1.fillna(data1.mean())
        data = pd.concat([data0, data1], axis=0)

2.划分训练集、测试集及数据不平衡处理

这里存在两个问题:
1.是否用整体的数据集先提取特征,再应用特征,然后划分训练集测试集?
这样可以较好的考虑到数据集的整体特征,但同时,测试集的表现可能会高于真实场景,考虑到数据集较大,故没有使用该方法。
2.数据平衡处理、数据集划分顺序
两种做法:先整体数据集平衡处理,再划分训练集、测试集 先划分训练集、测试集,再单独对训练集进行平衡处理(,若测试集样本比例极不平衡,也需要平衡处理)。
同样,由于第一种方法测试集表现可能过优,采用后一种。
这里,将两种方法进行实现,balance_prior=True,为第一种,否则第二种。

        # # 数据不平衡处理
        X, Y = data.drop([label_name], axis=1), data[label_name]
        if balance_prior:
            #先数据平衡处理,再拆分训练集、测试集
            smo = BorderlineSMOTE(kind='borderline-1',
                                  sampling_strategy={0: Y.value_counts()[0], 1: int(Y.value_counts()[0] * all_data_balance)},
                                  random_state=2)
            x, y = smo.fit_resample(X, Y)
            print("data balance result:\n", pd.Series(y).value_counts())
            # 打乱不平衡后的数据
            data_select = pd.concat((x, y), axis=1)
            data_select = shuffle(data_select,random_state=1)
            X, Y = data_select.drop([label_name], axis=1), data_select[label_name]

            #划分训练集、测试集
            X_train, X_test, y_train, y_test = model_selection.train_test_split(X,Y, test_size=0.3, random_state=1234)
        else:
            # 先拆分训练集、测试集,再分别对训练集、测试集平衡处理
            X_train, X_test, y_train, y_test = model_selection.train_test_split(X, Y, test_size=0.3, random_state=1234)

            # # 数据不平衡处理
            smo = BorderlineSMOTE(kind='borderline-1',
                                  sampling_strategy={0: y_train.value_counts()[0],
                                                     1: int(y_train.value_counts()[0] * train_data_balance)},
                                  random_state=2)
            x_train, y_train = smo.fit_resample(X_train, y_train)
            print("train data balance result:\n", pd.Series(y_train).value_counts())
            if test_data_balance is not None:
                smo = BorderlineSMOTE(kind='borderline-1',
                                      sampling_strategy={0: y_test.value_counts()[0],
                                                         1: int(y_test.value_counts()[0] * test_data_balance)},
                                      random_state=2)
                X_test, y_test = smo.fit_resample(X_test, y_test)
            print("test data balance result:\n", pd.Series(y_test).value_counts())
            # # 打乱平衡处理后的数据
            data_select = pd.concat((x_train, y_train), axis=1)
            data_select = shuffle(data_select, random_state=1)
            X_train, y_train = data_select.drop([label_name], axis=1), data_select[label_name]

        print("data train:\n", pd.Series(y_train).value_counts())
        print("data test:\n", pd.Series(y_test).value_counts())

3.数据标准化

        # 用训练集的数据特征对测试集标准化
        x_train= pd.DataFrame(self.coder.fit_transform(X_train.values))
        x_test =pd.DataFrame(self.coder.transform(X_test.values))

4.特征选择

在样本极其不平衡时,先将数据平衡处理再进行特征选择,选择的特征比较稳定。

1)随机森林

        # 随机森林
        if forest_n!=None:
            forest = RandomForestClassifier(random_state=1)
            forest.fit(x_train, y_train)
            ra=pd.Series(forest.feature_importances_,index=x_train.columns)
            self.forest_select= ra.sort_values()[::-1][:forest_n].index
            print("forest importance:",forest.feature_importances_)

            x_train = x_train.loc[:, self.forest_select]
            x_test = x_test.loc[:, self.forest_select]
            self.final_feature=self.final_feature[self.forest_select]
            print("forest features:", len(self.final_feature), "\n", self.final_feature)

2)相关性检测


        # 相关性检测
        if kbest_n != None:
            pearsonr_way = lambda x, y: np.array(list(map(lambda x: pearsonr(np.squeeze(x), np.squeeze(y)), x.T))).T[0]
            self.pearsonr_select = SelectKBest(pearsonr_way, k=kbest_n).fit(x_train, y_train).get_support(indices=True)
      
            x_train = x_train.loc[:, self.pearsonr_select]
            x_test = x_test.loc[:, self.pearsonr_select]
            self.final_feature = self.final_feature[self.pearsonr_select]
            print("kbest feature:",len(self.final_feature ),"\n", self.final_feature )

3)lasso特征

        # lasso特征
        if lasso:
            # 交叉验证找出最好的lambdas
            Lambdas = np.logspace(-5, 2, 200)
            lasso_cv = LassoCV(alphas=Lambdas, normalize=True, cv=10, max_iter=10000)
            lasso_cv.fit(x_train,y_train)

            lasso = Lasso(alpha=lasso_cv.alpha_, normalize=True, max_iter=10000)
            lasso.fit(x_train,y_train)
            print("lasso importance:\n",lasso.coef_,"\n","lasso mse:",mean_squared_error(y_train, lasso.predict(x_train)))
            tmp = pd.Series(index=x_train.columns.tolist(), data=lasso.coef_.tolist())
            self.lasso_select = x_train.columns[tmp.values != 0]

            x_train = x_train.loc[:, self.lasso_select]
            x_test= x_test.loc[:,self.lasso_select]
            self.final_feature = self.final_feature[self.lasso_select]
            print("lasso select:", len(self.final_feature), "\n", self.final_feature)

5.特征降维

使用PCA主成分分析

        # 主成分分析降维
        if pca_feature!=None:
            self.pca = PCA(n_components=pca_feature)
            self.pca.fit(x_train)
            print("pca importance:", self.pca.explained_variance_ratio_)
            x_train = self.pca.transform(x_train)
            x_test=self.pca.transform(x_test)
            
        return x_train, y_train,x_test,y_test

二、训练模型

1.训练

    def Train(self, file_path='', label_name="", col_miss_rate=0.5, row_miss_rate=0.8,
              balance_prior=False, all_data_balance=0.33, train_data_balance=1, test_data_balance=None,
              forest_n=20, kbest_n=None, lasso=False, pca_feature=None, ):

        x_train, y_train,x_test,y_test = self.preTrainDataAndSelectFeature(file_path=file_path, label_name=label_name,
                    col_miss_rate=col_miss_rate,row_miss_rate=row_miss_rate,
                    balance_prior=balance_prior, all_data_balance=all_data_balance,train_data_balance=train_data_balance, test_data_balance=test_data_balance,
                    forest_n=forest_n,kbest_n=kbest_n,lasso=lasso, pca_feature=pca_feature)

        self.svm.fit(x_train, y_train)

2.查看ROC曲线

用predict_proba()进行预测,得到各类别概率值

		self._showROC(self.svm.predict_proba(x_test)[:, 1], y_test)
    def _showROC(self, y_pred, y_true):
        fpr, tpr, thresholds = roc_curve(y_true,y_pred)
        roc_auc = auc(fpr, tpr)
        print("AUC:",roc_auc)

        plt.plot(fpr, tpr, 'b', label='Val AUC = %0.3f' % roc_auc)
        plt.legend(loc='upper right')
        plt.ylabel('True Positive Rate')
        plt.xlabel('False Positive Rate')
        plt.show()

3.评价指标

        y_pre=self.svm.predict(x_test)
        print("recall_score:", recall_score(y_test, y_pre))  
        print("precision_score:", precision_score(y_test, y_pre))  
        print("f1_score:", f1_score(y_test, y_pre))

三、测试数据预处理

1.去除训练集缺失率高的列

根据训练集的处理步骤,使其在标准化前拥有相同的特征

    def preEvalData(self,file_path="",label_name=""):
        x= pd.read_excel(file_path)
        #填充,若填充后仍有NAN,对x.mean()进行填充.fillna(0)
        x = x.fillna(x.mean().fillna(0))
        #去缺失特征和标签
        x=x.loc[:, self.drop_select]
        x = x.drop([label_name], axis=1)

2.标准化

        # 用训练集的均值和方差进行标准化
        x= pd.DataFrame(self.coder.transform(x.values))

3.特征选择

        if self.forest_select is not None:
            x = x.loc[:, self.forest_select]
        if self.lasso_select is not None:
            x = x.loc[:, self.lasso_select]
        if self.pearsonr_select is not None:
            x = x.loc[:, self.pearsonr_select]

4.特征降维

        if self.pca is not None:
            x = self.pca.transform(x.values)
            return x

四、测试集预测

    def predict(self, file_path='', label_name=""):
        x=self.preEvalData(file_path,label_name=label_name)

        pred=self.svm.predict_proba(x)

        count_1 = 0
        for i in pred[:,1]:
            #观察roc曲线得阈值为0.8
            if i >0.8 :
                count_1 = count_1 + 1

        print("1:", count_1)
        print("0:", len(pred) - count_1)

五、完整代码

# coding=utf-8
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from imblearn.over_sampling import BorderlineSMOTE
from sklearn.preprocessing import StandardScaler
from sklearn.utils import shuffle
from sklearn.feature_selection import SelectKBest
from sklearn.linear_model import LassoCV,Lasso
from sklearn.ensemble import RandomForestClassifier
from sklearn.decomposition import PCA
from sklearn import svm,model_selection
from sklearn.metrics import mean_squared_error,roc_curve, auc,recall_score, precision_score, f1_score, roc_curve,accuracy_score
from scipy.stats import pearsonr


class ML():
    def __init__(self):
        #特征、数据处理参数,用preTrainDataAndSelectFeature()计算出实际值
        #数据标准化模型,将训练集的均值方差应用到测试集
        self.coder=StandardScaler()
        #依据列缺失特征选择
        self.drop_select=None
        #随机森林特征选择
        self.forest_select=None
        #lasso回归特征选择
        self.lasso_select=None
        #kbest相关性检测特征选择
        self.pearsonr_select=None
        #pca数据降维模型
        self.pca= None
        # svc支持向量机模型
        self.svm = svm.SVC(C=0.8, kernel='rbf', gamma=20, decision_function_shape='ovr', probability=True)  
        #合并特征,为中文信息、非索引
        self.final_feature=None

    def preTrainDataAndSelectFeature(self, file_path="", label_name='', col_miss_rate=0.5,row_miss_rate=0.8,
                                     balance_prior=False,all_data_balance=0.33,train_data_balance=0.33,test_data_balance=None,
                                     forest_n=None, kbest_n=None, lasso=False, pca_feature=4, ):
        """
        训练集数据处理
        :param file_path:string 文件路径
        :param label_name:标签列名
        :param col_miss_rate:float 列缺失比率
        :param row_miss_rate:float 标签为0的行缺失率
        :param balance_prior:bool 数据集平衡优先
        :param all_data_balance:balance_prior=True时,float 整体数据不均衡调整比例 若=1,为0,1标签比值为1:1,=0.33为1:0.33
        :param train_data_balance:balance_prior=False时,
        :param test_data_balance:balance_prior=False时,
        :param forest_n:int 使用forest选取的特征数
        :param kbest_n:int 使用kbest选取的特征数
        :param lasso:bool 是否使用lasso回归选取特征
        :param pca_feature:int pca降维后的特征数(这是降维的最后一步,数据此后一直使用降维后的特征)
        :return:可直接放入模型训练的最终数据
        """
        data = pd.read_excel(file_path)

        # 对数据缺失进行处理
        data = data.dropna(axis=1, thresh=data.shape[0] * col_miss_rate)
        # 保存过滤缺失后的特征
        self.drop_select = list(data)
        self.final_feature = np.array(self.drop_select)
        print("drop select:",len(self.final_feature),"\n", self.final_feature)

        # 行处理:根据标签取值拆分样本,不同取值用各自的均值填充
        data_group = data.groupby(label_name)
        data0 = dict([x for x in data_group])[0]
        data0 = data0.dropna(axis=0, thresh=data0.shape[1] * row_miss_rate)
        data0 = data0.fillna(data0.mean())
        data1 = dict([x for x in data_group])[1]
        data1 = data1.fillna(data1.mean())
        data = pd.concat([data0, data1], axis=0)

        # # 数据不平衡处理
        X, Y = data.drop([label_name], axis=1), data[label_name]
        if balance_prior:
            #先数据平衡处理,再拆分训练集、测试集
            smo = BorderlineSMOTE(kind='borderline-1',
                                  sampling_strategy={0: Y.value_counts()[0], 1: int(Y.value_counts()[0] * all_data_balance)},
                                  random_state=2)
            x, y = smo.fit_resample(X, Y)
            print("data balance result:\n", pd.Series(y).value_counts())
            # 打乱不平衡后的数据
            data_select = pd.concat((x, y), axis=1)
            data_select = shuffle(data_select,random_state=1)
            X, Y = data_select.drop([label_name], axis=1), data_select[label_name]

            #划分训练集、测试集
            X_train, X_test, y_train, y_test = model_selection.train_test_split(X,Y, test_size=0.3, random_state=1234)
        else:
            # 先拆分训练集、测试集,再分别对训练集、测试集平衡处理
            X_train, X_test, y_train, y_test = model_selection.train_test_split(X, Y, test_size=0.3, random_state=1234)

            # # 数据不平衡处理
            smo = BorderlineSMOTE(kind='borderline-1',
                                  sampling_strategy={0: y_train.value_counts()[0],
                                                     1: int(y_train.value_counts()[0] * train_data_balance)},
                                  random_state=2)
            x_train, y_train = smo.fit_resample(X_train, y_train)
            print("train data balance result:\n", pd.Series(y_train).value_counts())
            if test_data_balance is not None:
                smo = BorderlineSMOTE(kind='borderline-1',
                                      sampling_strategy={0: y_test.value_counts()[0],
                                                         1: int(y_test.value_counts()[0] * test_data_balance)},
                                      random_state=2)
                X_test, y_test = smo.fit_resample(X_test, y_test)
            print("test data balance result:\n", pd.Series(y_test).value_counts())
            # # 打乱平衡处理后的数据
            data_select = pd.concat((x_train, y_train), axis=1)
            data_select = shuffle(data_select, random_state=1)
            X_train, y_train = data_select.drop([label_name], axis=1), data_select[label_name]

        print("data train:\n", pd.Series(y_train).value_counts())
        print("data test:\n", pd.Series(y_test).value_counts())

        # 用训练集的数据特征对测试集标准化
        x_train= pd.DataFrame(self.coder.fit_transform(X_train.values))
        x_test =pd.DataFrame(self.coder.transform(X_test.values))

        # 随机森林
        if forest_n!=None:
            forest = RandomForestClassifier(random_state=1)
            forest.fit(x_train, y_train)
            ra=pd.Series(forest.feature_importances_,index=x_train.columns)
            self.forest_select= ra.sort_values()[::-1][:forest_n].index
            print("forest importance:",forest.feature_importances_)

            x_train = x_train.loc[:, self.forest_select]
            x_test = x_test.loc[:, self.forest_select]
            self.final_feature=self.final_feature[self.forest_select]
            print("forest features:", len(self.final_feature), "\n", self.final_feature)

        # 相关性检测
        if kbest_n != None:
            pearsonr_way = lambda x, y: np.array(list(map(lambda x: pearsonr(np.squeeze(x), np.squeeze(y)), x.T))).T[0]
            self.pearsonr_select = SelectKBest(pearsonr_way, k=kbest_n).fit(x_train, y_train).get_support(indices=True)

            x_train = x_train.loc[:, self.pearsonr_select]
            x_test = x_test.loc[:, self.pearsonr_select]
            self.final_feature = self.final_feature[self.pearsonr_select]
            print("kbest feature:",len(self.final_feature ),"\n", self.final_feature )

        # lasso特征
        if lasso:
            # 交叉验证找出最好的lambdas
            Lambdas = np.logspace(-5, 2, 200)
            lasso_cv = LassoCV(alphas=Lambdas, normalize=True, cv=10, max_iter=10000)
            lasso_cv.fit(x_train,y_train)

            lasso = Lasso(alpha=lasso_cv.alpha_, normalize=True, max_iter=10000)
            lasso.fit(x_train,y_train)
            print("lasso importance:\n",lasso.coef_,"\n","lasso mse:",mean_squared_error(y_train, lasso.predict(x_train)))
            tmp = pd.Series(index=x_train.columns.tolist(), data=lasso.coef_.tolist())
            self.lasso_select = x_train.columns[tmp.values != 0]

            x_train = x_train.loc[:, self.lasso_select]
            x_test= x_test.loc[:,self.lasso_select]
            self.final_feature = self.final_feature[self.lasso_select]
            print("lasso select:", len(self.final_feature), "\n", self.final_feature)

        # 主成分分析降维
        if pca_feature!=None:
            self.pca = PCA(n_components=pca_feature)
            self.pca.fit(x_train)
            print("pca importance:", self.pca.explained_variance_ratio_)
            x_train = self.pca.transform(x_train)
            x_test=self.pca.transform(x_test)

        return x_train, y_train,x_test,y_test

    def preEvalData(self,file_path="",label_name=""):
        x = pd.read_excel(file_path)

        #填充,若填充后仍有NAN,对x.mean()进行填充.fillna(0)
        x = x.fillna(x.mean().fillna(0))
        #去缺失特征和标签
        x=x.loc[:, self.drop_select]
        x = x.drop([label_name], axis=1)
        # 用训练集的均值和方差进行标准化
        x= pd.DataFrame(self.coder.transform(x.values))
        if self.forest_select is not None:
            x = x.loc[:, self.forest_select]
        if self.lasso_select is not None:
            x = x.loc[:, self.lasso_select]
        if self.pearsonr_select is not None:
            x = x.loc[:, self.pearsonr_select]

        if self.pca is not None:
            x = self.pca.transform(x.values)
            return x
        return x.values

    def Train(self, file_path='', label_name="", col_miss_rate=0.5, row_miss_rate=0.8,
              balance_prior=False, all_data_balance=0.33, train_data_balance=1, test_data_balance=None,
              forest_n=20, kbest_n=None, lasso=False, pca_feature=None, ):

        x_train, y_train,x_test,y_test = self.preTrainDataAndSelectFeature(file_path=file_path, label_name=label_name,
                    col_miss_rate=col_miss_rate,row_miss_rate=row_miss_rate,
                    balance_prior=balance_prior, all_data_balance=all_data_balance,train_data_balance=train_data_balance, test_data_balance=test_data_balance,
                    forest_n=forest_n,kbest_n=kbest_n,lasso=lasso, pca_feature=pca_feature)

        self.svm.fit(x_train, y_train)
        self._showROC(self.svm.predict_proba(x_test)[:, 1], y_test)

        y_pre=self.svm.predict(x_test)
        print("recall_score:", recall_score(y_test, y_pre))
        print("precision_score:", precision_score(y_test, y_pre))
        print("f1_score:", f1_score(y_test, y_pre))

    def predict(self, file_path='', label_name=""):
        x=self.preEvalData(file_path,label_name=label_name)

        pred=self.svm.predict_proba(x)

        count_1 = 0
        for i in pred[:,1]:
            #观察roc曲线得阈值为0.8
            if i >0.8 :
                count_1 = count_1 + 1

        print("1:", count_1)
        print("0:", len(pred) - count_1)

    def _showROC(self, y_pred, y_true):
        fpr, tpr, thresholds = roc_curve(y_true,y_pred)
        roc_auc = auc(fpr, tpr)
        print("AUC:",roc_auc)

        plt.plot(fpr, tpr, 'b', label='Val AUC = %0.3f' % roc_auc)
        plt.legend(loc='upper right')
        plt.ylabel('True Positive Rate')
        plt.xlabel('False Positive Rate')
        plt.show()


if __name__=="__main__":
    pred=ML()

    pred.Train(file_path='.\xxxxxx.xlsx', label_name='是否真实', col_miss_rate=0.5, row_miss_rate=0.8,
               balance_prior=False, all_data_balance=None, train_data_balance=0.33, test_data_balance=0.1,
               forest_n=None, kbest_n=None, lasso=True, pca_feature=6, )

    #用训练好的模型预测
    pred.predict(file_path=r'xxxx.xlsx', label_name='是否真实')

六、问题归纳

1.数据集划分、数据平衡处理的顺序

2.数据标准化应该放在哪一步

3.关键参数

示例参数

col_miss_rate=0.2, row_miss_rate=0.65,
balance_prior=False, all_data_balance=None, train_data_balance=0.33, test_data_balance=None,
forest_n=None, kbest_n=None, lasso=True, pca_feature=5,

1)行列缺失率row_miss_rate、col_miss_rate

较大可能错过关键特征,较小可能包含没有价值的数据,对结果影响很大

2)模型参数

C,gamma等

3)PCA保留的维度数

pca_feature

4.roc曲线随测试集样本比例而变化,各种指标表现过低

1)检查每次运行筛选的特征是否相同,降维的模型参数是否相同,这是ROC变化过大的主要原因,很多模型具有随机性,每次运行可能会选出不同的特征,可设置random_state属性解决。
2)测试集数据样本极其不平衡(但不到万不得已,最好不改变测试集)
如下,样本比例小于10:1时,roc变化较小,指标稳定
当测试集标签比例0:1约为130:1时:
ROC曲线:
机器学习——数据处理流程梳理_第1张图片
AUC: 0.7369069343065693
recall_score: 0.2
precision_score: 0.04938271604938271
f1_score: 0.07920792079207921

50:1
机器学习——数据处理流程梳理_第2张图片
AUC: 0.8855932996443946
recall_score: 0.6153846153846154
precision_score: 0.3448275862068966
f1_score: 0.44198895027624313

10:1
机器学习——数据处理流程梳理_第3张图片
AUC: 0.9421023789389352
recall_score: 0.8353658536585366
precision_score: 0.7806267806267806
f1_score: 0.8070692194403534

3:1
机器学习——数据处理流程梳理_第4张图片
AUC: 0.9422224401968188
recall_score: 0.8407707910750507
precision_score: 0.9129955947136564
f1_score: 0.8753959873284055

你可能感兴趣的:(机器学习or深度学习,机器学习,python,数据分析,深度学习,人工智能)