def prune(L, minsup, k):
for name, pattern in L.items():
if pattern.support() < minsup:
del L[name]
else:
timeintervals, evids = pattern.timeintervals()
if len(timeintervals) > 0:
km.build(timeintervals, k)
# remove unsatisfied evidence
kluster = km.getClusters()
for e, c in zip(evids, km.cluster):
if len(kluster[c]) < minsup:
for key, value in e.items():
# print "yeah remove %s from %s" % (value, key)
pattern.removeEvidence(key, value)
if pattern.support() < minsup:
# print "prune by cluster"
del L[name]
def main():
data = spatio_data_generator(territory={'x':1000,'y':1000}, seeds=[{'x':200,'y':200}, {'x':700,'y':700}, {'x':700,'y':400}], pts = [50, 50, 50], noise=30, eps=100)
# x = []
# y = []
# start = time.time()
# rt = rtree.RTree(1, 10)
# for instance in data:
# rt.insert(rtree.Box(instance, [0,0]))
# end = time.time()
# timer = end-start
# print rt.root
# print "build tree completed in %.2fs" % timer
#
# query = rt.search(rtree.Box([200, 200], [100, 100]))
# print len(query)
# print query
#
# start = time.time()
# opt.build(data, 200, 10)
# end = time.time()
#
# print "===== Finished in %.2fs=====" % (end-start)
#
# n = opt.cluster(200)
# opt.report()
# opt.plot()
# opt.plotc(n)
# print n
# opt.plotr(400)
km.build(data, 3, 10, 1000)
print km.cluster
# km.plot()
clusters = km.getClusters()
print "%d clusters: %d and %d and %d" % (len(clusters), len(clusters[0]), len(clusters[1]), len(clusters[2]))
km.plot()