机器学习实验：朴素贝叶斯算法实现对iris鸢尾花数据集的分类

机器学习实验：朴素贝叶斯算法实现对iris鸢尾花数据集的分类

实验内容：

1. 通过sklearn提供的贝叶斯算法模型实现对iris鸢尾花数据集的分类及预测
2. 使用PCA降维，并通过matplotlib对决策边进行绘制

代码如下：

1. 导入相关库

   import pandas as pd
   import numpy as np
   import matplotlib.colors as colors
   import matplotlib as mpl
   import matplotlib.pyplot as plt
   import sklearn.metrics as metrics
   from sklearn.preprocessing import label_binarize
   from sklearn.datasets import load_iris
   from sklearn.datasets import load_wine
   from sklearn.datasets import load_breast_cancer
   from sklearn.model_selection import train_test_split
   import sklearn.naive_bayes as bayes
   from sklearn.decomposition import PCA
   
   mpl.rcParams['font.sans-serif']=[u'simHei']
   mpl.rcParams['axes.unicode_minus']=False
   

2. 导入数据集，并对数据集进行处理

   iris = load_iris()  #iris数据集
   X_iris=iris.data
   Y_iris=iris.target
   train_x_iris,test_x_iris,train_y_iris,test_y_iris=train_test_split(X_iris,Y_iris,test_size=0.3,random_state=2)
   # print(train_x_iris.shape)   #(105, 4)
   # print(test_x_iris.shape)    #(45, 4)
   

3. 实现绘图函数

   def drawing(train_x,test_x,train_y,test_y,Name='Gaussian',N=500):
       if(Name=='Gaussian'):
           model = bayes.GaussianNB()
       elif(Name == 'Bernoulli'):
           model = bayes.BernoulliNB()
       elif(Name == 'Multinomial'):
           model = bayes.MultinomialNB()
       else:
           model = bayes.GaussianNB()  #默认
       pca=PCA(n_components=2) #降为二维
   	train_x_iris=pca.fit_transform(train_x_iris,train_y_iris)   #(105, 2) 
   	test_x_iris=pca.fit_transform(test_x_iris,test_y_iris)      #(45, 2)
       model.fit(train_x,train_y)
       
       min_train_x1,min_train_x2=np.min(train_x,axis=0)
       max_train_x1,max_train_x2=np.max(train_x,axis=0)
       min_test_x1,min_test_x2=np.min(test_x,axis=0)
       max_test_x1,max_test_x2=np.max(test_x,axis=0)
       
       x1min=min(min_train_x1,min_test_x1)
       x1max=max(max_train_x1,max_test_x1)
       x1=np.linspace(x1min,x1max,N)
   
       x2min=min(min_train_x2,min_test_x2)
       x2max=max(max_train_x2,max_test_x2)
       x2=np.linspace(x2min,x2max,N)
   
       x1,x2=np.meshgrid(x1,x2)
   
       X=np.c_[x1.ravel(),x2.ravel()]
   
       y_hat=model.predict(X)
   
       plt.figure()
       plt.subplot(2,1,1)
       
       cmp=colors.ListedColormap([u'indigo', u'gold', u'hotpink', u'firebrick', u'indianred'])
   
       plt.pcolormesh(x1,x2,y_hat.reshape(x1.shape),shading='auto',cmap=cmp)
   
       plt.scatter(train_x[:,0],train_x[:,1],c=train_y)
       plt.scatter(test_x[:,0],test_x[:,1],c=test_y)
   
       plt.title(Name+":"+str(model.score(test_x,test_y)))
       plt.subplot(2,1,2)
       
       label=label_binarize(test_y,classes=(0,1,2))
   
       fpr,tpr,threofor=metrics.roc_curve(label.ravel(),model.predict_proba(test_x).ravel())
       plt.plot(fpr,tpr,c="red",label="ROC")
       plt.title("AUC"+str(metrics.auc(fpr,tpr)))
       plt.legend()
       plt.tight_layout()
       plt.show()
   

4. 对数据集进行分类及预测

   #1.获取朴素贝叶斯分类器的高斯模型,并训练
   gaussian = bayes.GaussianNB()
   bernoulli = bayes.BernoulliNB()
   multionmial =bayes.MultinomialNB()
   
   gaussian.fit(train_x_iris,train_y_iris)
   bernoulli.fit(train_x_iris,train_y_iris)
   multionmial.fit(train_x_iris,train_y_iris)    #输入数据出现负值,不能使用MultinomialNB
   
   #2.进行预测
   test_y_gaussianHat_iris = gaussian.predict(test_x_iris)
   test_y_bernoulliHat_iris = bernoulli.predict(test_x_iris)
   test_y_multionmialHat_iris = multionmial.predict(test_x_iris)
   
   #3.输出准确率
   score_gaussian = gaussian.score(test_x_iris,test_y_iris)
   score_bernoulli = bernoulli.score(test_x_iris,test_y_iris)
   score_multionmial = multionmial.score(test_x_iris,test_y_iris)
   print("gaussian score:"+str(score_gaussian))
   print("bernoulli score:"+str(score_bernoulli))
   print("multionmial score:"+str(score_multionmial))
   # gaussian score:0.9777777777777777
   # bernoulli score:0.28888888888888886
   # multionmial score:0.9111111111111111
   

5. 实现绘图，并实现决策边的绘制

   # 绘图
   drawing(train_x_iris,test_x_iris,train_y_iris,test_y_iris,'Gaussian')
   drawing(train_x_iris,test_x_iris,train_y_iris,test_y_iris,'Bernoulli')
   # drawing(train_x_iris,test_x_iris,train_y_iris,test_y_iris,'Multinomial')	#报错：Negative values in data passed to MultinomialNB (input X)，可能原因：数据降维后出现负值不适合于Multinomial模型
   

   结果如下：