python学习learning_curve和validation_curve

learning_curve是展示不同数据量，算法学习得分

validation_curve是展示某个因子，不同取值的算法得分

# -*- coding: utf-8 -*-"""Created on Wed Sep 06 09:55:13 2017@author: 飘的心"""from sklearn.model_selection import learning_curve #加载学习曲线from sklearn.model_selection import validation_curve #加载验证曲线from sklearn.model_selection import ShuffleSplit #加载数据处理from sklearn import datasets  #加载数据包from sklearn.naive_bayes import GaussianNB#加载高斯贝叶斯from sklearn.svm import LinearSVC #加载支持向量机import numpy as npimport matplotlib.pyplot as pltdata=datasets.load_digits()x=data.datay=data.targetcv=ShuffleSplit(test_size=0.25,random_state=0,n_splits=4) #处理数据，测试数据比例为0.25，4——折交叉estimator=GaussianNB()estimator.get_params().keys()  #可以获得学习算法参数train_sizes=[0.1,0.2,0.4,0.6,0.7,0.8]train_size,train_scores,test_scores=learning_curve(estimator,x,y,cv=cv,\                                    train_sizes=[0.1,0.2,0.4,0.6,0.7,0.8]) #获得学习曲线，针对不同的数据集new_train_scores=train_scores.mean(1)train_std=train_scores.std()test_std=test_scores.std()new_test_scores=test_scores.mean(1)'''画出不同比例数据集的学习曲线'''plt.grid()plt.fill_between(train_sizes,new_train_scores-train_std,                 new_train_scores+train_std,color='r',alpha=0.1)plt.fill_between(train_sizes,new_test_scores-test_std,                 new_test_scores+test_std,color='g',alpha=0.1)plt.plot(train_sizes,new_train_scores,'*-',c='r',label='train score')plt.plot(train_sizes,new_test_scores,'*-',c='g',label='test score')plt.legend(loc='best')plt.show()'''使用支持向量机，来做验证曲线'''estimator2=LinearSVC()estimator2.get_params().keys()#查看有哪些系数train_score2,validation_score2=validation_curve(estimator2,x,y,param_name='C',cv=cv                ,param_range=np.linspace(0.1,1,20)) #改变变量C，来看得分x_axis=np.linspace(0.1,1,20)train_score2_mean=train_score2.mean(1)train_score2_std=train_score2.std(1)validation_score2_mean=validation_score2.mean(1)validation_score2_std=validation_score2.std(1)plt.grid()plt.fill_between(x_axis,train_score2_mean-train_score2_std,                 train_score2_mean+train_score2_std,color='r',alpha=0.1)plt.fill_between(x_axis,validation_score2_mean-validation_score2_std,                 validation_score2_mean+validation_score2_std,color='g',alpha=0.1)plt.plot(x_axis,train_score2_mean,'o-',c='r',label='train score')plt.plot(x_axis,validation_score2_mean,'o-',c='g',label='validation score')plt.legend(loc='best')plt.show()