KNN with python

#!/usr/bin/python

"""
Created on Jan 13 , 2013
Filename: kNN.py
@author : libo
"""

from numpy import *
from sys import argv
from os import listdir

#scritp , filename = argv

def importData():
group = array([[1.0 , 1.1 ],[1.0 , 1.0],[0,0],[0,0.1]])
labels = ['A' , 'A' , 'B' , 'B']
return group , labels

def classify(inX , trainSet , labels , k):
dataSize = trainSet.shape[0]
diffMat = tile(inX , (dataSize ,1)) - trainSet
sqDiff = diffMat**2
sqDistances = sqDiff.sum(axis = 1)
distances= sqDistances**0.5
sortedIndex = argsort(distances)
classCount = {}
for index in range(k):
xlabel = labels[sortedIndex[index]]
classCount[xlabel] = classCount.get(xlabel, 0) + 1
#print classCount
sortedClassCount = sorted(classCount.iteritems() ,
key = lambda x:x[1] , reverse = True)
#print sortedClassCount
#print "result is : " , sortedClassCount[0][0]
return sortedClassCount[0][0]

def importData_M(filename):
fr = open(filename)
numOfLines = len(fr.readlines())
dataMat = zeros((numOfLines , 3))
classLabels = []
fr = open(filename)
index = 0
for line in fr.readlines():
line = line.strip()
lineList = line.split('\t')
dataMat[index , :] = lineList[0:3]
classLabels.append(lineList[-1])
index = index +1
return dataMat , classLabels

def autoNorm(dataSet):
#print dataSet
maxVals = dataSet.max(0)
minVals = dataSet.min(0)
ranges = maxVals - minVals
m = dataSet.shape[0]
normData = zeros(shape(dataSet))
diffData = dataSet - tile(minVals , (m , 1))
#normData = tile(ranges , (m , 1))
normedData = diffData/tile(ranges , (m ,1))
#print normedData
return normedData

def classTest():
hoRatio = 0.1
dataMat , dataLabels = importData_M("datingTestSet.txt")
normedMat = autoNorm(dataMat)
m = normedMat.shape[0]
testNum = int(m * hoRatio)
errorCount = 0.0
for i in range(testNum):
classifyLabel = classify(normedMat[i,:] , normedMat[testNum :m ,:] , dataLabels[testNum : m] , 5)
print "the classify come back with %r , the real labels is %r " %(classifyLabel , dataLabels[i])
if (classifyLabel != dataLabels[i]) :
errorCount += 1.0
print "the error rate is: %f" %(errorCount/float(testNum))

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

def handWriteTest():
hwLabels = []
trainingList = listdir("trainingDigits")
m = len(trainingList)
trainingMat = zeros((m , 1024))
for i in range(m ):
filenameStr = trainingList[i]
fileStr = filenameStr.split(".")[0]
classLabel = int(fileStr.split("_")[0])
hwLabels.append(classLabel)
trainingMat[i,:] = imgVector("trainingDigits/%s"\
%filenameStr)
testFileList = listdir("testDigits")
errorCount = 0.0
mTest = len(testFileList)
for i in range(mTest):
testFile = testFileList[i]
testStr = testFile.split(".")[0]
testLabel = int(testStr.split("_")[0])
testVector = imgVector("testDigits/%s" \
%testFile)
classResult = classify(testVector ,trainingMat , \
hwLabels , 3)
print "classify come back :%d , the real answer is :%d\
" %(classResult , testLabel)
if ( classResult != testLabel):
errorCount +=1
print "the total error num is : %d" %errorCount
print "the error rate is : %f" %(errorCount/float(mTest))

if __name__ == '__main__':
#	group , labels = importData()
#	group , labels = importData_M("datingTestSet.txt")
#	group = autoNorm(group)
#	classify([0,0 , 0] , group , labels , 3)
#	classTest()
handWriteTest()