机器学习模型评估与超参数调优详解

机器学习分为两类基本问题----回归与分类。在以前的文章中，也介绍了不少基本的机器学习模型。可在 Datawhale机器学习专辑 中 查看。 可是，当咱们创建好了相关模型之后咱们怎么评价咱们创建的模型的好坏以及优化咱们创建的模型呢？那本次分享的内容就是关于机器学习模型评估与超参数调优的。本次分享的内容包括：

• 用管道简化工做流

• 使用k折交叉验证评估模型性能

• 使用学习和验证曲线调试算法

• 经过网格搜索进行超参数调优

• 比较不一样的性能评估指标

1、用管道简化工做流

在不少机器学习算法中，咱们可能须要作一系列的基本操做后才能进行建模，如：在创建逻辑回归以前，咱们可能须要先对数据进行标准化，而后使用PCA将维，最后拟合逻辑回归模型并预测。那有没有什么办法能够同时进行这些操做，使得这些操做造成一个工做流呢？下面请看代码：

1. 加载基本工具库

import numpy as npimport pandas as pdimport matplotlib.pyplot as plt%matplotlib inlineplt.style.use("ggplot")import warningswarnings.filterwarnings("ignore")

2. 加载数据，并作基本预处理

# 加载数据df = pd.read_csv("http://archive.ics.uci.edu/ml/machine-learning-databases/breast-cancer-wisconsin/wdbc.data",header=None)# 作基本的数据预处理from sklearn.preprocessing import LabelEncoder
X = df.iloc[:,2:].valuesy = df.iloc[:,1].valuesle = LabelEncoder()    #将M-B等字符串编码成计算机能识别的0-1y = le.fit_transform(y)le.transform(['M','B'])# 数据切分8：2from sklearn.model_selection import train_test_split
X_train,X_test,y_train,y_test = train_test_split(X,y,test_size=0.2,stratify=y,random_state=1)

3. 把全部的操做所有封在一个管道pipeline内造成一个工做流：标准化+PCA+逻辑回归

完成以上操做，共有两种方式：

方式1：make_pipeline

# 把全部的操做所有封在一个管道pipeline内造成一个工做流：## 标准化+PCA+逻辑回归
### 方式1：make_pipelinefrom sklearn.preprocessing import StandardScalerfrom sklearn.decomposition import PCAfrom sklearn.linear_model import LogisticRegressionfrom sklearn.pipeline import make_pipeline
pipe_lr1 = make_pipeline(StandardScaler(),PCA(n_components=2),LogisticRegression(random_state=1))pipe_lr1.fit(X_train,y_train)y_pred1 = pipe_lr.predict(X_test)print("Test Accuracy: %.3f"% pipe_lr1.score(X_test,y_test))

Test Accuracy: 0.956

方式2：Pipeline

# 把全部的操做所有封在一个管道pipeline内造成一个工做流：## 标准化+PCA+逻辑回归
### 方式2：Pipelinefrom sklearn.preprocessing import StandardScalerfrom sklearn.decomposition import PCAfrom sklearn.linear_model import LogisticRegressionfrom sklearn.pipeline import Pipeline
pipe_lr2 = Pipeline([['std',StandardScaler()],['pca',PCA(n_components=2)],['lr',LogisticRegression(random_state=1)]])pipe_lr2.fit(X_train,y_train)y_pred2 = pipe_lr2.predict(X_test)print("Test Accuracy: %.3f"% pipe_lr2.score(X_test,y_test))

Test Accuracy: 0.956

2、使用k折交叉验证评估模型性能

评估方式1：k折交叉验证

• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  

# 评估方式1：k折交叉验证
from sklearn.model_selection import cross_val_score
scores1 = cross_val_score(estimator=pipe_lr,X = X_train,y = y_train,cv=10,n_jobs=1)print("CV accuracy scores:%s" % scores1)print("CV accuracy:%.3f +/-%.3f"%(np.mean(scores1),np.std(scores1)))

评估方式2：分层k折交叉验证

# 评估方式2：分层k折交叉验证
from sklearn.model_selection import StratifiedKFold
kfold = StratifiedKFold(n_splits=10,random_state=1).split(X_train,y_train)scores2 = []for k,(train,test) in enumerate(kfold):    pipe_lr.fit(X_train[train],y_train[train])    score = pipe_lr.score(X_train[test],y_train[test])    scores2.append(score)    print('Fold:%2d,Class dist.:%s,Acc:%.3f'%(k+1,np.bincount(y_train[train]),score))print('\nCV accuracy :%.3f +/-%.3f'%(np.mean(scores2),np.std(scores2)))

3、 使用学习和验证曲线调试算法

若是模型过于复杂，即模型有太多的自由度或者参数，就会有过拟合的风险（高方差）；而模型过于简单，则会有欠拟合的风险(高误差)。

下面咱们用这些曲线去识别并解决方差和误差问题：

1. 用学习曲线诊断误差与方差

# 用学习曲线诊断误差与方差from sklearn.model_selection import learning_curve
pipe_lr3 = make_pipeline(StandardScaler(),LogisticRegression(random_state=1,penalty='l2'))train_sizes,train_scores,test_scores = learning_curve(estimator=pipe_lr3,X=X_train,y=y_train,train_sizes=np.linspace(0.1,1,10),cv=10,n_jobs=1)train_mean = np.mean(train_scores,axis=1)train_std = np.std(train_scores,axis=1)test_mean = np.mean(test_scores,axis=1)test_std = np.std(test_scores,axis=1)plt.plot(train_sizes,train_mean,color='blue',marker='o',markersize=5,label='training accuracy')plt.fill_between(train_sizes,train_mean+train_std,train_mean-train_std,alpha=0.15,color='blue')plt.plot(train_sizes,test_mean,color='red',marker='s',markersize=5,label='validation accuracy')plt.fill_between(train_sizes,test_mean+test_std,test_mean-test_std,alpha=0.15,color='red')plt.xlabel("Number of training samples")plt.ylabel("Accuracy")plt.legend(loc='lower right')plt.ylim([0.8,1.02])plt.show()

2. 用验证曲线解决欠拟合和过拟合

# 用验证曲线解决欠拟合和过拟合from sklearn.model_selection import validation_curve
pipe_lr3 = make_pipeline(StandardScaler(),LogisticRegression(random_state=1,penalty='l2'))param_range = [0.001,0.01,0.1,1.0,10.0,100.0]train_scores,test_scores = validation_curve(estimator=pipe_lr3,X=X_train,y=y_train,param_name='logisticregression__C',param_range=param_range,cv=10,n_jobs=1)train_mean = np.mean(train_scores,axis=1)train_std = np.std(train_scores,axis=1)test_mean = np.mean(test_scores,axis=1)test_std = np.std(test_scores,axis=1)plt.plot(param_range,train_mean,color='blue',marker='o',markersize=5,label='training accuracy')plt.fill_between(param_range,train_mean+train_std,train_mean-train_std,alpha=0.15,color='blue')plt.plot(param_range,test_mean,color='red',marker='s',markersize=5,label='validation accuracy')plt.fill_between(param_range,test_mean+test_std,test_mean-test_std,alpha=0.15,color='red')plt.xscale('log')plt.xlabel("Parameter C")plt.ylabel("Accuracy")plt.legend(loc='lower right')plt.ylim([0.8,1.02])plt.show()

4、经过网格搜索进行超参数调优

若是只有一个参数须要调整，那么用验证曲线手动调整是一个好方法，可是随着须要调整的超参数愈来愈多的时候，咱们能不能自动去调整呢？！！！注意对比各个算法的时间复杂度。

（注意参数与超参数的区别：参数能够经过优化算法进行优化，如逻辑回归的系数；超参数是不能用优化模型进行优化的，如正则话的系数。）

方式1：网格搜索GridSearchCV()

# 方式1：网格搜索GridSearchCV()from sklearn.model_selection import GridSearchCVfrom sklearn.svm import SVCimport time
start_time = time.time()pipe_svc = make_pipeline(StandardScaler(),SVC(random_state=1))param_range = [0.0001,0.001,0.01,0.1,1.0,10.0,100.0,1000.0]param_grid = [{'svc__C':param_range,'svc__kernel':['linear']},{'svc__C':param_range,'svc__gamma':param_range,'svc__kernel':['rbf']}]gs = GridSearchCV(estimator=pipe_svc,param_grid=param_grid,scoring='accuracy',cv=10,n_jobs=-1)gs = gs.fit(X_train,y_train)end_time = time.time()print("网格搜索经历时间：%.3f S" % float(end_time-start_time))print(gs.best_score_)print(gs.best_params_)

方式2：随机网格搜索RandomizedSearchCV()

# 方式2：随机网格搜索RandomizedSearchCV()from sklearn.model_selection import RandomizedSearchCVfrom sklearn.svm import SVCimport time
start_time = time.time()pipe_svc = make_pipeline(StandardScaler(),SVC(random_state=1))param_range = [0.0001,0.001,0.01,0.1,1.0,10.0,100.0,1000.0]param_grid = [{'svc__C':param_range,'svc__kernel':['linear']},{'svc__C':param_range,'svc__gamma':param_range,'svc__kernel':['rbf']}]# param_grid = [{'svc__C':param_range,'svc__kernel':['linear','rbf'],'svc__gamma':param_range}]gs = RandomizedSearchCV(estimator=pipe_svc, param_distributions=param_grid,scoring='accuracy',cv=10,n_jobs=-1)gs = gs.fit(X_train,y_train)end_time = time.time()print("随机网格搜索经历时间：%.3f S" % float(end_time-start_time))print(gs.best_score_)print(gs.best_params_)

方式3 ：嵌套交叉验证

# 方式3：嵌套交叉验证from sklearn.model_selection import GridSearchCVfrom sklearn.svm import SVCfrom sklearn.model_selection import cross_val_scoreimport time
start_time = time.time()pipe_svc = make_pipeline(StandardScaler(),SVC(random_state=1))param_range = [0.0001,0.001,0.01,0.1,1.0,10.0,100.0,1000.0]param_grid = [{'svc__C':param_range,'svc__kernel':['linear']},{'svc__C':param_range,'svc__gamma':param_range,'svc__kernel':['rbf']}]gs = GridSearchCV(estimator=pipe_svc, param_grid=param_grid,scoring='accuracy',cv=2,n_jobs=-1)scores = cross_val_score(gs,X_train,y_train,scoring='accuracy',cv=5)end_time = time.time()print("嵌套交叉验证：%.3f S" % float(end_time-start_time))print('CV accuracy :%.3f +/-%.3f'%(np.mean(scores),np.std(scores)))

5、比较不一样的性能评估指标

有时候，准确率不是咱们惟一须要考虑的评价指标，由于有时候会存在各种预测错误的代价不同。例如：在预测一我的的肿瘤疾病的时候，若是病人A真实得肿瘤可是咱们预测他是没有肿瘤，跟A真实是健康可是预测他是肿瘤，两者付出的代价很大区别（想一想为何）。因此咱们须要其余更加普遍的指标：

1. 绘制混淆矩阵

# 绘制混淆矩阵from sklearn.metrics import confusion_matrix
pipe_svc.fit(X_train,y_train)y_pred = pipe_svc.predict(X_test)confmat = confusion_matrix(y_true=y_test,y_pred=y_pred)fig,ax = plt.subplots(figsize=(2.5,2.5))ax.matshow(confmat, cmap=plt.cm.Blues,alpha=0.3)for i in range(confmat.shape[0]):    for j in range(confmat.shape[1]):        ax.text(x=j,y=i,s=confmat[i,j],va='center',ha='center')plt.xlabel('predicted label')plt.ylabel('true label')plt.show()

2. 各类指标的计算

• 
  
• 
  
• 
  
• 
  
• 
  
• 
  

# 各类指标的计算from sklearn.metrics import precision_score,recall_score,f1_score
print('Precision:%.3f'%precision_score(y_true=y_test,y_pred=y_pred))print('recall_score:%.3f'%recall_score(y_true=y_test,y_pred=y_pred))print('f1_score:%.3f'%f1_score(y_true=y_test,y_pred=y_pred))

3. 将不一样的指标与GridSearch结合

• 
  
• 
  
• 
  
• 
  
• 
  
• 
  
• 
  

# 将不一样的指标与GridSearch结合from sklearn.metrics import make_scorer,f1_scorescorer = make_scorer(f1_score,pos_label=0)gs = GridSearchCV(estimator=pipe_svc,param_grid=param_grid,scoring=scorer,cv=10)gs = gs.fit(X_train,y_train)print(gs.best_score_)print(gs.best_params_)

4. 绘制ROC曲线

# 绘制ROC曲线from sklearn.metrics import roc_curve,aucfrom sklearn.metrics import make_scorer,f1_scorescorer = make_scorer(f1_score,pos_label=0)gs = GridSearchCV(estimator=pipe_svc,param_grid=param_grid,scoring=scorer,cv=10)y_pred = gs.fit(X_train,y_train).decision_function(X_test)#y_pred = gs.predict(X_test)fpr,tpr,threshold = roc_curve(y_test, y_pred) ###计算真阳率和假阳率roc_auc = auc(fpr,tpr) ###计算auc的值plt.figure()lw = 2plt.figure(figsize=(7,5))plt.plot(fpr, tpr, color='darkorange',         lw=lw, label='ROC curve (area = %0.2f)' % roc_auc) ###假阳率为横坐标，真阳率为纵坐标作曲线plt.plot([0, 1], [0, 1], color='navy', lw=lw, linestyle='--')plt.xlim([-0.05, 1.0])plt.ylim([-0.05, 1.05])plt.xlabel('False Positive Rate')plt.ylabel('True Positive Rate')plt.title('Receiver operating characteristic ')plt.legend(loc="lower right")plt.show()