def kMeans(data, nrOFCluster, nrOfGivenIterations):
centroids = findCentroids(nrOFCluster, data)
listD = [[0 for i in range(0, nrOFCluster)], [0 for i in range(0, nrOFCluster)]]
loopCounter = 0
givenError = 0.5
while loopCounter < nrOfGivenIterations:
assignment = assignToCluster(nrOFCluster, data, centroids)
# seeking for empty centroids
empty = []
for e in range(0, nrOFCluster):
if assignment.count(e) == 0:
empty.append(e)
# print "empty ", empty, "\n\n"
#collecting empty centroids
emptyCentroid = Dataset([])
emptyListD = [[], []]
for q in range(0, empty.__len__()):
emptyCentroid.getListOfVectors().append(centroids.getListOfVectors()[empty[q]])
emptyListD[0].append(listD[0][empty[q]])
emptyListD[1].append(listD[1][empty[q]])
#calculation
D = meanQuantizationError(nrOFCluster, data, assignment, centroids)
centroids = calculateNewCentroids(nrOFCluster, data, assignment, centroids)
numberOfExistingCluster = []
if (loopCounter == 0):
listD[0] = D
else:
if loopCounter == 1:
listD[1] = D
else:
listD[0] = listD[1]
listD[1] = D
# check if all cluster have at least one Vector assigned
# otherwise del from cluster, decrement nrOFCluster
centroidsTmp = Dataset([])
listDTmp= [[],[]]
for z in range(0, centroids.getListOfVectors().__len__()):
ctr = 0
for q in range(0, centroids.getListOfVectors()[0].__len__()):
if centroids.getListOfVectors()[z][q] != 0:
ctr += 1
if ctr == 0:
centroidsTmp.getListOfVectors().append(centroids.getListOfVectors()[z])
listDTmp[0].append(listD[0][z])
listDTmp[1].append(listD[1][z])
else:
numberOfExistingCluster.append(z)
#remove empty data from processing
for q in range(0,centroidsTmp.getListOfVectors().__len__()):
centroids.getListOfVectors().remove(centroidsTmp.getListOfVectors()[q])
listD[0].remove(listDTmp[0][q])
listD[1].remove(listDTmp[1][q])
nrOFCluster -= 1
#check if given error tha calculated
QoS = 0
for q in range(0, numberOfExistingCluster.__len__()):
if countDiff(listD[0][q], listD[1][q]) < givenError:
QoS += 1
if QoS == centroids.getListOfVectors().__len__():
print "\n!!!!!!!!!!KONIEC!!!!!!!!!!"
break
#restore empty centroids
centroids.getListOfVectors().extend(emptyCentroid.getListOfVectors())
nrOFCluster += emptyCentroid.getListOfVectors().__len__()
listD[1].extend(emptyListD[0])
listD[0].extend(emptyListD[1])
print "@@ centr= " , centroids.getListOfVectors()
print "\n*******END OF LOOP " ,loopCounter, "\n"
loopCounter += 1
for i in range(len(assignment)):
print assignment[i], getAllFiles('../input')[i]
if QoS == centroids.getListOfVectors().__len__():
print "\n!!!!!!!!!!KONIEC!!!!!!!!!!"
break
#restore empty centroids
centroids.getListOfVectors().extend(emptyCentroid.getListOfVectors())
nrOFCluster += emptyCentroid.getListOfVectors().__len__()
listD[1].extend(emptyListD[0])
listD[0].extend(emptyListD[1])
print "@@ centr= " , centroids.getListOfVectors()
print "\n*******END OF LOOP " ,loopCounter, "\n"
loopCounter += 1
for i in range(len(assignment)):
print assignment[i], getAllFiles('../input')[i]
nrOFCluster = 3
nrOfGivenIterations = 3
data = Dataset(prepareInput('../input', 10, '/home/kchrusci/Workspace/repo/k-means/projektpython/ForbiddenWords.txt'))
print "data = ", data.getListOfVectors()
kMeans(data,nrOFCluster,nrOfGivenIterations)
print(getAllFiles('/home/koper/PycharmProjects/First/Samples'))
