def subComparison(folder1, folder2, stringDiffParam, bagWordParam, apiCallParam, dancingBunnyParam):
min = float('inf')
clust = -1
foundClust = []
for file1 in os.listdir(folder1):
currClust = 0
for file2 in os.listdir(folder2):
dist = fileComparison(folder1 + file1, folder2 + file2,
stringDiffParam, bagWordParam, apiCallParam, dancingBunnyParam)
if dist < min:
min = dist
clust = currClust
currClust += 1
if clust not in foundClust:
foundClust.append(clust)
print "Number of clusters: " + str(len(foundClust))
return foundClust
def cluster():
m = numpy.array([[0 for i in rangeFiles] for j in rangeFiles])
for i in rangeFiles:
for j in range(i + 1, numFiles):
distance = float(
fileComparison(
folder + filesToCluster[i],
folder + filesToCluster[j],
stringDiffParam,
bagWordParam,
apiCallParam,
dancingBunnyParam,
)
)
# Square to minimize quadratic error
# distanceSquared = distance * distance
m[i][j] = distance
m[j][i] = distance
print "Distances matrix calculated"
print "Median of matrix: " + str(numpy.median(m))
labels, clusters = kmedoids(m)
print "Labels:"
print labels
print "Clusters:"
print clusters
print "K-medoids run, number of clusters: " + str(len(clusters))
moveFiles(filesToCluster, labels, clusters, "KMedoids/")
# Negate since affinity propagation works on negative distance values
m = -m
gc.collect()
labels, clusters = affinityPropagation(m)
print "Affinity propagation run, number of clusters: " + str(len(clusters))
moveFiles(filesToCluster, labels, clusters, "AffinityPropagation/")
