def reCluster(clusterList):
'''
Resort the clusters according to which centroid is closest
Param: clusterList - a list of every cluster to be reclustered
Return: clusterList - the resorted list of clusters
avg - the average distance of a tag and it's new cluster after resorting
'''
iterCount = 0
avg = 0
for cluster in clusterList:
cluster.wipeMembers()
tagList = TAG_DICT.keys()
for tag in tagList:
vector = TAG_DICT[tag]
matchTuple = sorted(clusterMatch(clusterList, vector),
key=lambda tup: tup[1],
reverse=False)[0]
if matchTuple[0] > 0:
bestIndex = matchTuple[0]
avg += matchTuple[1]
clusterList[bestIndex].addMember(tag, vector)
iterCount += 1
# if iterCount % 500 == 0:
# print(matchTuple[1], tag, matchTuple[0])
avg = avg / iterCount
for cluster in clusterList:
cluster.setCentroid()
return clusterList, avg
def reCluster(clusterList): ''' Resort the clusters according to which centroid is closest Param: clusterList - a list of every cluster to be reclustered Return: clusterList - the resorted list of clusters avg - the average distance of a tag and it's new cluster after resorting ''' iterCount = 0 avg = 0 for cluster in clusterList: cluster.wipeMembers() tagList = TAG_DICT.keys() for tag in tagList: vector = TAG_DICT[tag] matchTuple = sorted(clusterMatch(clusterList, vector), key=lambda tup: tup[1], reverse=False)[0] if matchTuple[0] > 0: bestIndex = matchTuple[0] avg += matchTuple[1] clusterList[bestIndex].addMember(tag, vector) iterCount +=1 # if iterCount % 500 == 0: # print(matchTuple[1], tag, matchTuple[0]) avg = avg/iterCount for cluster in clusterList: cluster.setCentroid() return clusterList, avg
def randomCluster(numClusters, tagDict):
'''
Set up for recursive clustering, randomly distributing urls across n clusters
Param: numClusters - the number of clusters to randomly distribute tag vectors across
Return: clusteredUrls - a list of size numClusters of TagCluster objects
'''
clusterList = [tumblruser.TagCluster() for x in range(numClusters)]
for entry in tagDict:
loc = random.randint(0, numClusters - 1)
clusterList[loc].addMember(entry, tagDict[entry])
for cluster in clusterList:
cluster.setCentroid()
return clusterList
def randomCluster(numClusters, tagDict): ''' Set up for recursive clustering, randomly distributing urls across n clusters Param: numClusters - the number of clusters to randomly distribute tag vectors across Return: clusteredUrls - a list of size numClusters of TagCluster objects ''' clusterList = [tumblruser.TagCluster() for x in range(numClusters)] for entry in tagDict: loc = random.randint(0, numClusters - 1) clusterList[loc].addMember(entry, tagDict[entry]) for cluster in clusterList: cluster.setCentroid() return clusterList
