机器学习-KNN识别手写数字

逻辑：

代码

import os
from com.test.my_test03 import *

# 将每一个文件的内容拼接成一行
def img2vector(fileName):
    returnVect = np.zeros((1, 1024))
    file = open(fileName)
    for i in range(32):
        line = file.readline()
        if(line != "
"):
            for j in range(32):
                returnVect[0, 32*i +j] = int(line[j])
    return returnVect

class IdentifImgClassTest(object):
    # 每个文件的代表数字
    hwLabels = []
    # 获取TrainData目录下所有的文件
    trainingFileList = os.listdir('D:/code/python/test2/data/TrainData')
    # 训练集的数据量，空间中分布的数据量  m=1934
    m = len(trainingFileList)
    # 1934*1024的零矩阵
    trainingMat = np.zeros((m, 1024))
    # 构造训练集
    for i in range(m):
        fileName = trainingFileList[i]
        fileNameSplit = fileName.split('.')[0]
        num = int(fileNameSplit.split('_')[0])
        hwLabels.append(num)
        trainingMat[i, :] = img2vector('D:/code/python/test2/data/TrainData/%s' % fileName)

    #构造测试集
    testFileList = os.listdir('D:\code\python\test2\data\TestData')
    errorCount = 0.0
    mTest = len(testFileList)
    for i in range(mTest):
        fileNameStr = testFileList[i]
        fileStr = fileNameStr.split('.')[0]
        classNumStr = int(fileStr.split('_')[0])
        vectorUnderTest = img2vector('D:/code/python/test2/data/TestData/%s' % fileNameStr)
        # 将训练集数据和第一条测试数据 以及K值传递给classify这个方法
        classifierResult = classify(vectorUnderTest, trainingMat, hwLabels, 3)
        print("识别出的数字是: %d, 真实数字是: %d" % (classifierResult, classNumStr))
        if (classifierResult != classNumStr): errorCount += 1.0
    print("
识别错误次数 %d" % errorCount)
    errorRate = errorCount / float(mTest)
    print("
正确率: %f" % (1 - errorRate))

if __name__ == '__main__':
    IdentifImgClassTest()

结果：