k邻近算法识别手写数字

from numpy import *
from os import listdir
import operator


def distance(array_measured_tar_dist, array_dist, label_dist, k):
    lens_dist = array_dist.shape[0]
    array_tar = tile(array_measured_tar_dist, (lens_dist, 1)) - array_dist
    array_tar_sq = array_tar ** 2
    distance_sq = array_tar_sq.sum(axis=1)
    distances = distance_sq ** 0.5
    sorted_list = distances.argsort()
    class_count = {}
    for i in range(k):
        nor_label = label_dist[sorted_list[i]]
        class_count[nor_label] = class_count.get(nor_label, 0) + 1
    sorted_class_count = sorted(class_count.items(), key=operator.itemgetter(1), reverse=True)
    return sorted_class_count[0][0]


def img_vector(filename):
    return_vec = zeros((1, 1024))
    fr = open(filename)
    for i in range(32):
        line_str = fr.readline()
        for j in range(32):
            return_vec[0, i * 32 + j] = int(line_str[j])
    return return_vec


def hand_write_train():
    labels_hw = []
    train_filename_list_hw = listdir('trainingDigits')
    train_lens_hw = len(train_filename_list_hw)
    train_array = zeros((train_lens_hw, 1024))
    for i in range(train_lens_hw):
        filename_str = train_filename_list_hw[i]
        file_str = filename_str.split('.')[0]
        real_num = int(file_str.split('_')[0])
        labels_hw.append(real_num)
        train_array[i] = img_vector('trainingDigits/%s' % filename_str)
    test_filename_list_hw = listdir('testDigits')
    error_count = 0.0
    test_lens_hw = len(test_filename_list_hw)
    for i in range(test_lens_hw):
        filename_str = test_filename_list_hw[i]
        file_str = filename_str.split('.')[0]
        real_num = int(file_str.split('_')[0])
        test_array = img_vector('testDigits/%s' % filename_str)
        result_test = distance(test_array, train_array, labels_hw, 3)
        print('the tester came back with %d,the real answer is %d' % (result_test, real_num))
        if result_test != real_num:
            error_count += 1
    error_rate = float(error_count/test_lens_hw)
    print('the test number of error is %d' % error_count)
    print('thr test error rate is %f' % error_rate)


if __name__ == '__main__':
    # a = img_vector('testDigits/0_0.txt')
    # print(a)

    hand_write_train()

代码运行结果

...
the tester came back with 9,the real answer is 9
the tester came back with 9,the real answer is 9
the tester came back with 9,the real answer is 9
the tester came back with 9,the real answer is 9
the tester came back with 9,the real answer is 9
the test number of error is 10
thr test error rate is 0.010571

训练文件与测试文件已打包上传

https://pan.baidu.com/s/1ut_D7utHW1d57G7r15zRKw