# store the total time needed to compute distance matrix
matrixCompTotal = 0
# store the total time needed to compute modularity
modularityTotal = 0
# store the total time needed to split clusters
splitTotal = 0
# store the total time needed to compute cluster diameter
diaTotal = 0
# store the total time needed to find out feature exclusion
excluTotal = 0
if __name__ == '__main__':
ngramPath = sys.argv[1]
outPath = sys.argv[2]
sizeThreshold = float(sys.argv[3]) if len(sys.argv) > 3 else 0.05
startTime = time.time()
sid_seq = rhc.getSidNgramMap(ngramPath)
print('[LOG]: total users %d' % len(sid_seq))
result = run(ngramPath, sid_seq, outPath)
json.dump(result, open('%sresult.json' % outPath, 'w'))
print('[STAT]: total time %f' % (time.time() - startTime))
print(('[STAT]: maxtrix com: %f, dismeter: %f, modularity: %f, split: %f, '
'exclusion: %f') %
(matrixCompTotal, diaTotal, modularityTotal, splitTotal, excluTotal))
"""
import json
import sys
import numpy as np
import recursiveHierarchicalClustering as rhc
inPath = sys.argv[1]
sid_ngram = sys.argv[2]
outPath = sys.argv[3]
# display 24 bins in visulization
binCount = 24
data = json.load(open(inPath))
sid_seq = rhc.getSidNgramMap(sid_ngram)
def allUser(tree):
""" output all users in the tree """
users = []
if (tree[0] == 'l'):
return tree[1]
for subTree in tree[1]:
users.extend(allUser(subTree))
return users
def getPatternDist(pattern, sids, sid_seq):
"""
get pattern distribution