Question

我正在尝试为数据集构建svm分类器，我知道svm分类器可用于2D数组，但是此代码不起作用，因为程序将newtemp2视为3D数组，所以我想知道我必须为我的数据使用svm classifier做些什么。

    train_setfeat = []
    train_setlabel = []
    newtemp2=[]
    for vector in newtemp:
        newtemp2.append(np.reshape(vector, (431, 19)))
        #convert each vector to 2d array

    j = 0
    for vector in newtemp2:
        if j < 2100: # 70 % for train

            train_setfeat.append(vector)
            train_setlabel.append(classlabels[j])
            j += 1
        else:
            break


    test_setfeat = []
    test_setlabel = []
    j = 0
    for vector in newtemp2:
        if j < 2997 and j >= 2100:   #20 % for test
            test_setfeat.append(vector)
            test_setlabel.append(classlabels[j])
        if j>= 3000:
            break
        j += 1

    classifier1 = svm.SVC(kernel='linear')
    classifier1.fit(train_setfeat, train_setlabel)

#sample of newtemp data
newtemp =[
    (0.05,0.0,0.0,0.02,0.0),
    (0.0,0.0,0.0,0.02,0.0),
    (0.05,0.0,0.0)]

如果找到单词，则数据集中的每个句子都表示为矢量0.0，否则将单词的权重

Answer 1

在使用列表和numpy数组组合创建训练集时遇到一些问题。

尝试这部分代码，应通过使用以下代码替换第3-5行来解决您的问题：

N=len(newtemp)
newtemp2=np.empty(N,431,19)
i=0;
for vector in newtemp:
    newtemp2[i,:]=np.reshape(vector, (431, 19)))
    i+=1

您可以对其余代码执行相同的操作

如何使用SVM分类器进行分类？

1 个答案: