K-Nearest neighbor algorithm for machine learning Combat (handwriting recognition system)

This thing is similar to the one that improved the dating site, which is to convert all the numbers into black and white images of 32*32 pixels in advance, then convert them into character metalized (denoted by 0,1), save all 1024 pixels in a one-dimensional matrix so that Euclidean distances can be calculated by KNN to get the closest answer.

1 ImportOS2 Importoperator3  fromNumPyImport*4 5 defclassify0 (InX, DataSet, labels, k):6Datasetsize =Dataset.shape[0]7Diffmat = Tile (InX, (datasetsize,1))-DataSet#Unified matrix for Add and subtract8Sqdiffmat = diffmat**29Sqdistances = Sqdiffmat.sum (Axis=1)#to accumulate, axis=0 is by column, Axis=1 is by rowTendistances = sqdistances**0.5#Open Radicals OneSorteddistindicies = Distances.argsort ()#sort in ascending order, return the original subscript AClassCount = {} -      forIinchRange (k): -Voteilabel =Labels[sorteddistindicies[i]] theClasscount[voteilabel] = classcount.get (Voteilabel, 0) + 1#get is a method in the dictionary, preceded by the value to be obtained, followed by the default value if the value does not exist -Sortedclasscount = sorted (Classcount.items (), Key=operator.itemgetter (1), reverse=True) -     returnSortedclasscount[0][0] -  +  - defimg2vector (filename): +f =open (filename) AReturnvect = Zeros ((1,1024)) at      forIinchRange (32): -line =F.readline () -          forJinchRange (32): -RETURNVECT[0,I*32+J] =Int (line[j]) -     returnReturnvect -  in  - defhandwritingclasstest (): toFileList = Os.listdir ('trainingdigits') +m =Len (fileList) -Traingmat = Zeros ((M, 1024)) theHwlabels = [] *      forIinchRange (m): $FileName =Filelist[i]Panax Notoginsengprefix = Filename.split ('.') [0] -Number = Int (Prefix.split ('_') [0]) the hwlabels.append (number) +Traingmat[i,:] = Img2vector ('trainingdigits/%s'%fileName) ATestfilelist = Os.listdir ('testdigits') them =Len (testfilelist) +ErrorNum = 0.0 -      forIinchRange (m): $Testfilename =Testfilelist[i] $prefix = Testfilelist[i].split ('.') [0] -realnumber = Int (Prefix.split ('_') [0]) -Testmat = Img2vector ('testdigits/%s'%testfilename) theTestResult = Classify0 (Testmat, Traingmat, Hwlabels, 3) -         ifTestResult! =RealNumber:WuyiErrorNum + = 1 the         Print('The classifier came back with:%d, the real answer is:%d'%(TestResult, realnumber)) -     Print('error rate is%f'% (errornum/float (m))) Wu  - if __name__=='__main__': AboutHandwritingclasstest ()

K-Nearest neighbor algorithm for machine learning Combat (handwriting recognition system)