基于sklearn几种分类算法

基于sklearn的多元分类的svm

from data_test_kNN import file2matrix

from sklearn import svm

'''将文件数据转化为测试数据和训练数据'''

def get_trian_testdata(filename):

    dataMat,dataClassLabel=file2matrix(filename)

    return dataMat,dataClassLabel

'''训练svm模型'''

def train_svm(trainMat,trainClassLabel,decision_function_shape):

    '''kernel : string, optional (default='rbf')

         Specifies the kernel type to be used in the algorithm.

         It must be one of 'linear', 'poly', 'rbf', 'sigmoid', 'precomputed' or

         a callable.

         If none is given, 'rbf' will be used. If a callable is given it is

         used to pre-compute the kernel matrix from data matrices; that matrix

         should be an array of shape ``(n_samples, n_samples)``.

    '''

    clf=svm.SVC(decision_function_shape)

    clf.fit(trainMat,trainClassLabel)

    return clf

'''识别率'''

def testAcc(clf,testMat,testClassLabel):

    testAcc=clf.score(testMat,testClassLabel)

    return testAcc

'''预测数据'''

def predict(clf,datMat):

    res=clf.predict(datMat)

    return res  

'''主函数'''    

if __name__ == "__main__":

    trainMat,trainClassLabel=get_trian_testdata('trainingDigits/train.txt')

    testMat,testClassLabel=get_trian_testdata('testDigits/test.txt')

    print '训练数据集为：',len(trainClassLabel)
    print '测试数据集为：',len(testClassLabel)

   '''使用一对一的方式实现svm多元分类'''

    #核函数为rbf,c=1

    clf=svm.SVC(decision_function_shape='ovo')

    clf.fit(trainMat,trainClassLabel)

    test=clf.score(testMat,testClassLabel)

    print '核函数为rbf的识别率：',test

    #核函数为linear,c=1

    clf=svm.SVC(decision_function_shape='ovo',kernel='linear')

    clf.fit(trainMat,trainClassLabel)

    test=clf.score(testMat,testClassLabel)

    print '核函数为linear的识别率：',test

    #核函数为poly,c=1

    clf=svm.SVC(decision_function_shape='ovo',kernel='poly')

    clf.fit(trainMat,trainClassLabel)

    test=clf.score(testMat,testClassLabel)

    print '核函数为poly的识别率：',test

    #核函数为sigmoid,c=1

    clf=svm.SVC(decision_function_shape='ovo',kernel='sigmoid')

    clf.fit(trainMat,trainClassLabel)

    test=clf.score(testMat,testClassLabel)

    print '核函数为sigmoid的识别率：',test

    '''使用一对多的方式实现svm多元分类'''

    #核函数为rbf,c=1

    clf=svm.SVC(decision_function_shape='ovr')

    clf.fit(trainMat,trainClassLabel)

    test=clf.score(testMat,testClassLabel)

    print '核函数为rbf的识别率：',test

    #核函数为linear,c=1

    clf=svm.SVC(decision_function_shape='ovr',kernel='linear')

    clf.fit(trainMat,trainClassLabel)

    test=clf.score(testMat,testClassLabel)

    print '核函数为linear的识别率：',test

    #核函数为poly,c=1

    clf=svm.SVC(decision_function_shape='ovr',kernel='poly')

    clf.fit(trainMat,trainClassLabel)

    test=clf.score(testMat,testClassLabel)

    print '核函数为poly的识别率：',test

    #核函数为sigmoid,c=1

    clf=svm.SVC(decision_function_shape='ovr',kernel='sigmoid')

    clf.fit(trainMat,trainClassLabel)

    test=clf.score(testMat,testClassLabel)
    print '核函数为sigmoid的识别率：',test

基于sklearn的决策树

from data_test_kNN import file2matrix

from sklearn import tree

'''将文件数据转化为测试数据和训练数据'''

def get_trian_testdata(filename):

    dataMat,dataClassLabel=file2matrix(filename)

    return dataMat,dataClassLabel

'''

   max_depth : int or None, optional (default=None)

        The maximum depth of the tree. If None, then nodes are expanded until

        all leaves are pure or until all leaves contain less than

        min_samples_split samples（default=2）.

        Ignored if ``max_leaf_nodes`` is not None.

'''

'''主函数'''    

if __name__ == "__main__":

    trainMat,trainClassLabel=get_trian_testdata('trainingDigits/train.txt')

    testMat,testClassLabel=get_trian_testdata('testDigits/test.txt')

    print '训练数据集为：',len(trainClassLabel)

    print '测试数据集为：',len(testClassLabel)

    '''max_depth没被设置'''

    clf=tree.DecisionTreeClassifier()

    clf=clf.fit(trainMat,trainClassLabel)

    testAcc=clf.score(testMat,testClassLabel)

    print testAcc