版權(quán)聲明:本文為博主原創(chuàng)文章,轉(zhuǎn)載請(qǐng)指明轉(zhuǎn)載地址
www.cnblogs.com/fydeblog/p/7140974.html
前言
這篇notebook是關(guān)于機(jī)器學(xué)習(xí)中監(jiān)督學(xué)習(xí)的k近鄰算法,將介紹2個(gè)實(shí)例,分別是使用k-近鄰算法改進(jìn)約會(huì)網(wǎng)站的效果和手寫識(shí)別系統(tǒng).
操作系統(tǒng):ubuntu14.04 運(yùn)行環(huán)境:anaconda-python2.7-notebook 參考書籍:機(jī)器學(xué)習(xí)實(shí)戰(zhàn) notebook writer ----方陽(yáng)
k-近鄰算法(kNN)的工作原理:存在一個(gè)樣本數(shù)據(jù)集合,也稱作訓(xùn)練樣本集,并且樣本集中的每個(gè)數(shù)據(jù)都存在標(biāo)簽,即我們知道樣本集中每一組數(shù)據(jù)與所屬分類的對(duì)應(yīng)關(guān)系,輸入沒有標(biāo)簽的新數(shù)據(jù)后,將新數(shù)據(jù)的每個(gè)特征與樣本集中數(shù)據(jù)對(duì)應(yīng)的特征進(jìn)行比較,然后算法提取樣本集中特征最相似的分類標(biāo)簽。
注意事項(xiàng):在這里說一句,默認(rèn)環(huán)境python2.7的notebook,用python3.6的會(huì)出問題,還有我的目錄可能跟你們的不一樣,你們自己跑的時(shí)候記得改目錄,我會(huì)把notebook和代碼以及數(shù)據(jù)集放到結(jié)尾的百度云盤,方便你們下載!
1.改進(jìn)約會(huì)網(wǎng)站的匹配效果
1-1.準(zhǔn)備導(dǎo)入數(shù)據(jù)
1 from numpy import *2 import operator3 4 def createDataSet():5 group = array([[1.0,1.1],[1.0,1.0],[0,0],[0,0.1]])6 labels = ['A','A','B','B']7 return group, labels
先來點(diǎn)開胃菜,在上面的代碼中,我們導(dǎo)入了兩個(gè)模塊,一個(gè)是科學(xué)計(jì)算包numpy,一個(gè)是運(yùn)算符模塊,在后面都會(huì)用到,在createDataSet函數(shù)中,我們初始化了group,labels,我們將做這樣一件事,[1.0,1.1]和[1.0,1.0] 對(duì)應(yīng)屬于labels中 A 分類,[0,0]和[0,0.1]對(duì)應(yīng)屬于labels中的B分類,我們想輸入一個(gè)新的二維坐標(biāo),根據(jù)上面的坐標(biāo)來判斷新的坐標(biāo)屬于那一類,在這之前,我們要實(shí)現(xiàn)k-近鄰算法,下面就開始實(shí)現(xiàn)
1 def classify0(inX, dataSet, labels, k): 2 dataSetSize = dataSet.shape[0] 3 diffMat = tile(inX, (dataSetSize,1)) - dataSet 4 sqDiffMat = diffMat**2 5 sqDistances = sqDiffMat.sum(axis=1) 6 distances = sqDistances**0.5 7 sortedDistIndicies = distances.argsort() 8 classCount={} 9 for i in range(k):10 voteIlabel = labels[sortedDistIndicies[i]]11 classCount[voteIlabel] = classCount.get(voteIlabel,0) + 112 sortedClassCount = sorted(classCount.iteritems(), key=operator.itemgetter(1), reverse=True)13 return sortedClassCount[0][0]
代碼解析:
函數(shù)的第一行是要得到數(shù)據(jù)集的數(shù)目,例如group.shape就是(4,2),shape[0]反應(yīng)數(shù)據(jù)集的行,shape[1]反應(yīng)列數(shù)
函數(shù)的第二行是array對(duì)應(yīng)相減,tile會(huì)生成關(guān)于Inx的dataSetSize大小的array,例如,InX是[0,0],則tile(InX,(4,1))是array([[0, 0], [0, 0], [0, 0],[0, 0]]),然后與dataSet對(duì)應(yīng)相減,得到新的array
函數(shù)的第三行是對(duì)第二步的結(jié)果進(jìn)行平方算法,方便下一步算距離
函數(shù)的第四行是進(jìn)行求和,注意是axis=1,也就是array每個(gè)二維數(shù)組成員進(jìn)行求和(行求和),如果是axis=0就是列求和
第五行是進(jìn)行平方距離的開根號(hào)
以上5行實(shí)現(xiàn)的是距離的計(jì)算 ,下面的是選出距離最小的k個(gè)點(diǎn),對(duì)類別進(jìn)行統(tǒng)計(jì),返回所占數(shù)目多的類別
classCount定義為存儲(chǔ)字典,里面有‘A’和‘B’,它們的值是在前k個(gè)距離最小的數(shù)據(jù)集中的個(gè)數(shù),本例最后classCount={'A':1,'B':2},函數(shù)argsort是返回array數(shù)組從小到大的排列的序號(hào),get函數(shù)返回字典的鍵值,由于后面加了1,所以每次出現(xiàn)鍵值就加1,就可以就算出鍵值出現(xiàn)的次數(shù)里。最后通過sorted函數(shù)將classCount字典分解為列表,sorted函數(shù)的第二個(gè)參數(shù)導(dǎo)入了運(yùn)算符模塊的itemgetter方法,按照第二個(gè)元素的次序(即數(shù)字)進(jìn)行排序,由于此處reverse=True,是逆序,所以按照從大到小的次序排列。
1-2.準(zhǔn)備數(shù)據(jù):從文本中解析數(shù)據(jù)
這上面是k-近鄰的一個(gè)小例子,我的標(biāo)題還沒介紹,現(xiàn)在來介紹標(biāo)題,準(zhǔn)備數(shù)據(jù),一般都是從文本文件中解析數(shù)據(jù),還是從一個(gè)例子開始吧!
本次例子是改進(jìn)約會(huì)網(wǎng)站的效果,我們定義三個(gè)特征來判別三種類型的人
特征一:每年獲得的飛行??屠锍虜?shù)
特征二:玩視頻游戲所耗時(shí)間百分比
特征三:每周消費(fèi)的冰淇淋公升數(shù)
根據(jù)以上三個(gè)特征:來判斷一個(gè)人是否是自己不喜歡的人,還是魅力一般的人,還是極具魅力的人
于是,收集了1000個(gè)樣本,放在datingTestSet2.txt中,共有1000行,每一行有四列,前三列是特征,后三列是從屬那一類人,于是問題來了,我們這個(gè)文本文件的輸入導(dǎo)入到python中來處理,于是需要一個(gè)轉(zhuǎn)換函數(shù)file2matrix,函數(shù)輸入是文件名字字符串,輸出是訓(xùn)練樣本矩陣(特征矩陣)和類標(biāo)簽向量
1 def file2matrix(filename): 2 fr = open(filename) 3 numberOfLines = len(fr.readlines()) #get the number of lines in the file 4 returnMat = zeros((numberOfLines,3)) #prepare matrix to return 5 classLabelVector = [] #prepare labels return 6 fr = open(filename) 7 index = 0 8 for line in fr.readlines(): 9 line = line.strip()10 listFromLine = line.split('\t')11 returnMat[index,:] = listFromLine[0:3]12 classLabelVector.append(int(listFromLine[-1]))13 index += 114 return returnMat,classLabelVector
這個(gè)函數(shù)比較簡(jiǎn)單,就不詳細(xì)說明里,這里只介紹以下一些函數(shù)的功能吧!
open函數(shù)是打開文件,里面必須是字符串,由于后面沒加‘w’,所以是讀文件
readlines函數(shù)是一次讀完文件,通過len函數(shù)就得到文件的行數(shù)
zeros函數(shù)是生成numberOfLines X 3的矩陣,是array型的
strip函數(shù)是截掉所有的回車符
split函數(shù)是以輸入?yún)?shù)為分隔符,輸出分割后的數(shù)據(jù),本例是制表鍵,最后輸出元素列表
append函數(shù)是向列表中加入數(shù)據(jù)
1-3.分析數(shù)據(jù):使用Matplotlib創(chuàng)建散點(diǎn)圖
首先,從上一步得到訓(xùn)練樣本矩陣和類標(biāo)簽向量,先更換一下路徑
cd /home/fangyang/桌面/machinelearninginaction/Ch02/
datingDataMat,datingLabels = file2matrix('datingTestSet2.txt')
1 import matplotlib2 import matplotlib.pyplot as plt3 fig = plt.figure()4 ax = fig.add_subplot(111)5 ax.scatter(datingDataMat[:,0], datingDataMat[:,1], 15.0*array(datingLabels), 15.0*array(datingLabels)) #scatter函數(shù)是用來畫散點(diǎn)圖的6 plt.show()
結(jié)果顯示
1-4. 準(zhǔn)備數(shù)據(jù): 歸一化處理
我們從上圖可以上出,橫坐標(biāo)的特征值是遠(yuǎn)大于縱坐標(biāo)的特征值的,這樣再算新數(shù)據(jù)和數(shù)據(jù)集的數(shù)據(jù)的距離時(shí),數(shù)字差值最大的屬性對(duì)計(jì)算結(jié)果的影響最大,我們就可能會(huì)丟失掉其他屬性,例如這個(gè)例子,每年獲取的飛行??屠锍虜?shù)對(duì)計(jì)算結(jié)果的影響遠(yuǎn)大于其余兩個(gè)特征,這是我們不想看到的,所以這里采用歸一化數(shù)值處理,也叫特征縮放,用于將特征縮放到同一個(gè)范圍內(nèi)。
本例的縮放公式 newValue = (oldValue - min) / (max - min)
其中min和max是數(shù)據(jù)集中的最小特征值和最大特征值。通過該公式可將特征縮放到區(qū)間(0,1)
下面是特征縮放的代碼
1 def autoNorm(dataSet):2 minVals = dataSet.min(0)3 maxVals = dataSet.max(0)4 ranges = maxVals - minVals5 normDataSet = zeros(shape(dataSet))6 m = dataSet.shape[0]7 normDataSet = dataSet - tile(minVals, (m,1))8 normDataSet = normDataSet/tile(ranges, (m,1)) #element wise divide9 return normDataSet, ranges, minVals
normDataSet(1000 X 3)是歸一化后的數(shù)據(jù),range(1X3)是特征的范圍差(即最大值減去最小值),minVals(1X3)是最小值。
原理上面已介紹,這里不在復(fù)述。
1-5.測(cè)試算法:作為完整程序驗(yàn)證分類器
好了,我們已經(jīng)有了k-近鄰算法、從文本解析出數(shù)據(jù)、還有歸一化處理,現(xiàn)在可以使用之前的數(shù)據(jù)進(jìn)行測(cè)試了,測(cè)試代碼如下
1 def datingClassTest(): 2 hoRatio = 0.50 3 datingDataMat,datingLabels = file2matrix('datingTestSet2.txt') #load data setfrom file 4 normMat, ranges, minVals = autoNorm(datingDataMat) 5 m = normMat.shape[0] 6 numTestVecs = int(m*hoRatio) 7 errorCount = 0.0 8 for i in range(numTestVecs): 9 classifierResult = classify0(normMat[i,:],normMat[numTestVecs:m,:],datingLabels[numTestVecs:m],3)10 print "the classifier came back with: %d, the real answer is: %d" % (classifierResult, datingLabels[i])11 if (classifierResult != datingLabels[i]): errorCount += 1.012 print "the total error rate is: %f" % (errorCount/float(numTestVecs))13 print errorCount
這里函數(shù)用到里之前講的三個(gè)函數(shù):file2matrix、autoNorm和classify0.這個(gè)函數(shù)將數(shù)據(jù)集分成兩個(gè)部分,一部分當(dāng)作分類器的訓(xùn)練樣本,一部分當(dāng)作測(cè)試樣本,通過hoRatio進(jìn)行控制,函數(shù)hoRatio是0.5,它與樣本總數(shù)相乘,將數(shù)據(jù)集平分,如果想把訓(xùn)練樣本調(diào)大一些,可增大hoRatio,但最好不要超過0.8,以免測(cè)試樣本過少,在函數(shù)的最后,加了錯(cuò)誤累加部分,預(yù)測(cè)出來的結(jié)果不等于實(shí)際結(jié)果,errorCount就加1,然后最后除以總數(shù)就得到錯(cuò)誤的概率。
說了這么多,都還沒有測(cè)試以下,下面來測(cè)試一下!先從簡(jiǎn)單的開始(已將上面的函數(shù)放在kNN.py中了)
1 import kNN2 group , labels = kNN.createDataSet()
group #結(jié)果在下
array([[ 1. , 1.1], [ 1. , 1. ], [ 0. , 0. ], [ 0. , 0.1]])
labels #結(jié)果在下
['A', 'A', 'B', 'B']
這個(gè)小例子最開始提過,有兩個(gè)分類A和B,通過上面的group為訓(xùn)練樣本,測(cè)試新的數(shù)據(jù)屬于那一類
1 kNN.classify0([0,0], group, labels, 3) #使用k-近鄰算法進(jìn)行測(cè)試
'B' #結(jié)果是B分類
直觀地可以看出[0,0]是與B所在的樣本更近,下面來測(cè)試一下約會(huì)網(wǎng)站的匹配效果
先將文本中的數(shù)據(jù)導(dǎo)出來,由于前面在分析數(shù)據(jù)畫圖的時(shí)候已經(jīng)用到里file2matrix,這里就不重復(fù)用了。
datingDataMat #結(jié)果在下
array([[ 4.09200000e+04, 8.32697600e+00, 9.53952000e-01], [ 1.44880000e+04, 7.15346900e+00, 1.67390400e+00], [ 2.60520000e+04, 1.44187100e+00, 8.05124000e-01], ..., [ 2.65750000e+04, 1.06501020e+01, 8.66627000e-01], [ 4.81110000e+04, 9.13452800e+00, 7.28045000e-01], [ 4.37570000e+04, 7.88260100e+00, 1.33244600e+00]])
datingLabels #由于過長(zhǎng),只截取一部分,詳細(xì)去看jupyter notebook
然后對(duì)數(shù)據(jù)進(jìn)行歸一化處理
1 normMat , ranges , minVals = kNN.autoNorm(datingDataMat) #使用歸一化函數(shù)
normMat
array([[ 0.44832535, 0.39805139, 0.56233353], [ 0.15873259, 0.34195467, 0.98724416], [ 0.28542943, 0.06892523, 0.47449629], ..., [ 0.29115949, 0.50910294, 0.51079493], [ 0.52711097, 0.43665451, 0.4290048 ], [ 0.47940793, 0.3768091 , 0.78571804]])
ranges
array([ 9.12730000e+04, 2.09193490e+01, 1.69436100e+00])
minVals
array([ 0. , 0. , 0.001156])
最后進(jìn)行測(cè)試,運(yùn)行之前的測(cè)試函數(shù)datingClassTest
1 kNN.datingClassTest()
由于過長(zhǎng),只截取一部分,詳細(xì)去看jupyter notebook
可以看到上面結(jié)果出現(xiàn)錯(cuò)誤32個(gè),錯(cuò)誤率6.4%,所以這個(gè)系統(tǒng)還算不錯(cuò)!
1-6.系統(tǒng)實(shí)現(xiàn)
我們可以看到,測(cè)試固然不錯(cuò),但用戶交互式很差,所以結(jié)合上面,我們要寫一個(gè)完整的系統(tǒng),代碼如下:
1 def classifyPerson(): 2 resultList = ['not at all', 'in small doses', 'in large doses'] 3 percentTats = float(raw_input("percentage of time spent playing video games?")) 4 ffMiles = float(raw_input("frequent flier miles earned per year?")) 5 iceCream = float(raw_input("liters of ice cream consumed per year?")) 6 datingDataMat, datingLabels = file2matrix('datingTestSet2.txt') 7 normMat, ranges, minVals = autoNorm(datingDataMat) 8 inArr = array([ffMiles, percentTats, iceCream]) 9 classifierResult = classify0((inArr-minVals)/ranges, normMat, datingLabels,3)10 print "You will probably like this person" , resultList[classifierResult - 1]
運(yùn)行情況
1 kNN.classifyPerson()
percentage of time spent playing video games?10 #這里的數(shù)字都是用戶自己輸入的frequent flier miles earned per year?10000liters of ice cream consumed per year?0.5You will probably like this person in small doses
這個(gè)就是由用戶自己輸出參數(shù),并判斷出感興趣程度,非常友好
2. 手寫識(shí)別系統(tǒng)
下面再介紹一個(gè)例子,也是用k-近鄰算法,去實(shí)現(xiàn)對(duì)一個(gè)數(shù)字的判斷,首先我們是將寬高是32X32的像素的黑白圖像轉(zhuǎn)換成文本文件存儲(chǔ),但我們知道文本文件必須轉(zhuǎn)換成特征向量,才能進(jìn)入k-近鄰算法中進(jìn)行處理,所以我們需要一個(gè)img2vector函數(shù)去實(shí)現(xiàn)這個(gè)功能!
img2vector代碼如下:
1 def img2vector(filename):2 returnVect = zeros((1,1024))3 fr = open(filename)4 for i in range(32):5 lineStr = fr.readline()6 for j in range(32):7 returnVect[0,32*i+j] = int(lineStr[j])8 return returnVect
這個(gè)函數(shù)挺簡(jiǎn)單的,先用zeros生成1024的一維array,然后用兩重循環(huán),外循環(huán)以行遞進(jìn),內(nèi)循環(huán)以列遞進(jìn),將32X32的文本數(shù)據(jù)依次賦值給returnVect
好了,轉(zhuǎn)換函數(shù)寫好了,說一下訓(xùn)練集和測(cè)試集,所有的訓(xùn)練集都放在trainingDigits文件夾中,測(cè)試集放在testDigits文件夾中,訓(xùn)練集有兩千個(gè)樣本,0~9各有200個(gè),測(cè)試集大約有900個(gè)樣本,這里注意一點(diǎn),所有在文件夾里的命名方式是有要求的,我們是通過命名方式來解析出它的真實(shí)數(shù)字,然后與通過k-近鄰算法得出的結(jié)果相對(duì)比,例如945.txt,這里的數(shù)字是9,連接符前面的數(shù)字就是這個(gè)樣本的真實(shí)數(shù)據(jù)。該系統(tǒng)實(shí)現(xiàn)的方法與前面的約會(huì)網(wǎng)站的類似,就不多說了。
系統(tǒng)測(cè)試代碼如下
1 def handwritingClassTest(): 2 hwLabels = [] 3 trainingFileList = listdir('trainingDigits') #load the training set 4 m = len(trainingFileList) 5 trainingMat = zeros((m,1024)) 6 for i in range(m): 7 fileNameStr = trainingFileList[i] 8 fileStr = fileNameStr.split('.')[0] #take off .txt 9 classNumStr = int(fileStr.split('_')[0])10 hwLabels.append(classNumStr)11 trainingMat[i,:] = img2vector('trainingDigits/%s' % fileNameStr)12 testFileList = listdir('testDigits') #iterate through the test set13 errorCount = 0.014 mTest = len(testFileList)15 for i in range(mTest):16 fileNameStr = testFileList[i]17 fileStr = fileNameStr.split('.')[0] #take off .txt18 classNumStr = int(fileStr.split('_')[0])19 vectorUnderTest = img2vector('testDigits/%s' % fileNameStr)20 classifierResult = classify0(vectorUnderTest, trainingMat, hwLabels, 3)21 print "the classifier came back with: %d, the real answer is: %d" % (classifierResult, classNumStr)22 if (classifierResult != classNumStr): errorCount += 1.023 print "\nthe total number of errors is: %d" % errorCount24 print "\nthe total error rate is: %f" % (errorCount/float(mTest))
這里的listdir是從os模塊導(dǎo)入的,它的功能是列出給定目錄下的所有文件名,以字符串形式存放,輸出是一個(gè)列表
這里的split函數(shù)是要分離符號(hào),得到該文本的真實(shí)數(shù)據(jù),第一個(gè)split函數(shù)是以小數(shù)點(diǎn)為分隔符,例如‘1_186.txt’ ,就變成了['1_186','txt'],然后取出第一個(gè),就截掉了.txt,第二個(gè)split函數(shù)是以連接符_為分隔符,就截掉后面的序號(hào),剩下前面的字符數(shù)據(jù)‘1’,然后轉(zhuǎn)成int型就得到了它的真實(shí)數(shù)據(jù),其他的沒什么,跟前面一樣
下面開始測(cè)試
1 kNN.handwritingClassTest()
我們可以看到最后結(jié)果,錯(cuò)誤率1.2%, 可見效果還不錯(cuò)!
這里把整個(gè)kNN.py文件貼出來,主要是上面已經(jīng)介紹的函數(shù)
'''Input: inX: vector to compare to existing dataset (1xN) dataSet: size m data set of known vectors (NxM) labels: data set labels (1xM vector) k: number of neighbors to use for comparison (should be an odd number) Output: the most popular class label'''from numpy import *import operatorfrom os import listdirdef classify0(inX, dataSet, labels, k): dataSetSize = dataSet.shape[0] diffMat = tile(inX, (dataSetSize,1)) - dataSet sqDiffMat = diffMat**2 sqDistances = sqDiffMat.sum(axis=1) distances = sqDistances**0.5 sortedDistIndicies = distances.argsort() classCount={} for i in range(k): voteIlabel = labels[sortedDistIndicies[i]] classCount[voteIlabel] = classCount.get(voteIlabel,0) + 1 sortedClassCount = sorted(classCount.iteritems(), key=operator.itemgetter(1), reverse=True) return sortedClassCount[0][0]def createDataSet(): group = array([[1.0,1.1],[1.0,1.0],[0,0],[0,0.1]]) labels = ['A','A','B','B'] return group, labelsdef file2matrix(filename): fr = open(filename) numberOfLines = len(fr.readlines()) #get the number of lines in the file returnMat = zeros((numberOfLines,3)) #prepare matrix to return classLabelVector = [] #prepare labels return fr = open(filename) index = 0 for line in fr.readlines(): line = line.strip() listFromLine = line.split('\t') returnMat[index,:] = listFromLine[0:3] classLabelVector.append(int(listFromLine[-1])) index += 1 return returnMat,classLabelVector def autoNorm(dataSet): minVals = dataSet.min(0) maxVals = dataSet.max(0) ranges = maxVals - minVals normDataSet = zeros(shape(dataSet)) m = dataSet.shape[0] normDataSet = dataSet - tile(minVals, (m,1)) normDataSet = normDataSet/tile(ranges, (m,1)) #element wise divide return normDataSet, ranges, minVals def datingClassTest(): hoRatio = 0.50 #hold out 10% datingDataMat,datingLabels = file2matrix('datingTestSet2.txt') #load data setfrom file normMat, ranges, minVals = autoNorm(datingDataMat) m = normMat.shape[0] numTestVecs = int(m*hoRatio) errorCount = 0.0 for i in range(numTestVecs): classifierResult = classify0(normMat[i,:],normMat[numTestVecs:m,:],datingLabels[numTestVecs:m],3) print "the classifier came back with: %d, the real answer is: %d" % (classifierResult, datingLabels[i]) if (classifierResult != datingLabels[i]): errorCount += 1.0 print "the total error rate is: %f" % (errorCount/float(numTestVecs)) print errorCount def classifyPerson(): resultList = ['not at all', 'in small doses', 'in large doses'] percentTats = float(raw_input("percentage of time spent playing video games?")) ffMiles = float(raw_input("frequent flier miles earned per year?")) iceCream = float(raw_input("liters of ice cream consumed per year?")) datingDataMat, datingLabels = file2matrix('datingTestSet2.txt') normMat, ranges, minVals = autoNorm(datingDataMat) inArr = array([ffMiles, percentTats, iceCream]) classifierResult = classify0((inArr-minVals)/ranges, normMat, datingLabels,3) print "You will probably like this person" , resultList[classifierResult - 1] def img2vector(filename): returnVect = zeros((1,1024)) fr = open(filename) for i in range(32): lineStr = fr.readline() for j in range(32): returnVect[0,32*i+j] = int(lineStr[j]) return returnVectdef handwritingClassTest(): hwLabels = [] trainingFileList = listdir('trainingDigits') #load the training set m = len(trainingFileList) trainingMat = zeros((m,1024)) for i in range(m): fileNameStr = trainingFileList[i] fileStr = fileNameStr.split('.')[0] #take off .txt classNumStr = int(fileStr.split('_')[0]) hwLabels.append(classNumStr) trainingMat[i,:] = img2vector('trainingDigits/%s' % fileNameStr) testFileList = listdir('testDigits') #iterate through the test set errorCount = 0.0 mTest = len(testFileList) for i in range(mTest): fileNameStr = testFileList[i] fileStr = fileNameStr.split('.')[0] #take off .txt classNumStr = int(fileStr.split('_')[0]) vectorUnderTest = img2vector('testDigits/%s' % fileNameStr) classifierResult = classify0(vectorUnderTest, trainingMat, hwLabels, 3) print "the classifier came back with: %d, the real answer is: %d" % (classifierResult, classNumStr) if (classifierResult != classNumStr): errorCount += 1.0 print "\nthe total number of errors is: %d" % errorCount print "\nthe total error rate is: %f" % (errorCount/float(mTest))
結(jié)尾
至此,這個(gè)k-近鄰算法的介紹到這里就結(jié)束了,希望這篇文章對(duì)你的學(xué)習(xí)有幫助!
百度云鏈接: https://http://www.cnblogs.com/fydeblog/p/7140974.htmlpan.baidu.com/s/1mhLcAy0 密碼: e8dp
http://www.cnblogs.com/fydeblog/p/7140974.html