KNN算法实现（Mnist数据集）

参考 http://blog.csdn.net/wds2006sdo/article/details/51933044

#!/usr/bin/python# -*- coding:utf8 -*-import pandas as pdimport numpy as npimport timefrom sklearn.model_selection import train_test_splitfrom sklearn.metrics import accuracy_scoredef predict(test_features,train_features,train_labels):    predict=[]    count=0    train_size=len(train_labels)    test_size=len(test_features)    for test_vec in test_features:        print 'predicting:',count,'/',test_size        count+=1        max_index=-1        dist_max=0        knn_list = []        for i in range(k):            train_vec=train_features[i]            dist=np.sqrt(np.sum(np.square(test_vec-train_vec)))            label=train_labels[i]            knn_list.append((dist,label))        for i in range(k,train_size):            label = train_labels[i]            train_vec = train_features[i]            dist = np.sqrt(np.sum(np.square(test_vec - train_vec)))            #find the max value of present knn_list            if max_index==-1:             # 如果 max_index=-1,则发生了交换，需要重新寻找交换后 knn_list的最大值                for j in range(k):                    if knn_list[j][0] > dist_max:                        dist_max=knn_list[j][0]                        max_index=j            # if dist < dist_max,swap them            if dist < dist_max:                knn_list[max_index]=(dist,label)                dist_max=0        # dist_max=0 ,为下次寻找最大值                max_index=-1        #class_count=[0 for i in range(k)]        class_count=np.zeros(k)        for dist,label in knn_list:                  #notice            class_count[label]+=1        max_count=max(class_count)        for i in range(k):            if max_count==class_count[i]:                predict.append(i)              #notice append ()                break    return np.array(predict)k=10if __name__=='__main__':    print ('Start reading data...')    time_1 = time.time()    raw_data=pd.read_csv('../data/train.csv')    data=raw_data.values    imgs=data[::,1::]    labels=data[::,0]    train_features,test_features,train_labels,test_labels=train_test_split(imgs,labels,test_size=0.33,random_state=1)    time_2 = time.time()    print 'reading  cost ', time_2 - time_1, ' seconds', '\n'    print 'Start predicting...'    test_predict = predict(test_features, train_features, train_labels)    time_4 = time.time()    print 'predicting cost ', time_4 - time_2, ' seconds', '\n'    score = accuracy_score(test_labels, test_predict)    print "The accruacy socre is ", score