def run_diagnose(self):
standards = serial.objFromFile('prob_groups.txt')
c = ClusterBridge()
name_net = c.getNameNet()
for thres in range(1, 10):
# Experimental
def lengthFilter(triplet):
names = triplet[1]
return len(names) == thres
filtered_net = filter(lengthFilter, name_net)
# Experimental-end
observed = gstat.convertNameNet2Observed(filtered_net)
prediction = matchstat.convertObserved2Prediction(standards, observed)
# Format becomes:
# [(true1, true2,...), (predict1, predict2, ...), (meta1, meta2, ...)]
pairs = matchstat.unzipGroupPrediction(prediction)
num_label = 4
tester = tests.tester(num_label)
assert(len(pairs) >= 2) # sanity check
tester.record(pairs[0], pairs[1])
tester.recordMeta(pairs[2])
#print thres
print "Accuracy:", tester.accuracy()
print "Conf matrix:"
print tester.confusionMatrix(friends.groupToNumeric)
def test1():
# A list of standard distributions. Format is
# [(age, prob), (age, prob), ...]
standards = serial.objFromFile('prob_groups.txt')
# Age and a set of names. Format is
# [(age, [names...]), (age, [names...])]
name_net = friends.getNameNetworkSampled(10, 100)
# Format is:
# [(age, prob, meta), (age, prob, meta), ...]
observed = convertNameNet2Observed(name_net)
def match(o): # a single observation, i.e., (age, [names...])
age = o[0]
p = o[1]
meta = o[2]
if sum(p) != 0:
matched = matchObserved(standards, p)
matched_ages = map(lambda x: x[0], matched)
return (age, p, matched_ages, meta)
else:
return (age, 0, meta)
matched_arr = map(match, observed)
print serial.toString(matched_arr)
def simulateIdealFriendshipNetwork(n_friends):
standards = serial.objFromFile('prob_groups.txt')
#standards = gstat.generateEmpiricalDistributionFromSample(500, 20)
name_net = friends.getPseudoNameNetSampled(n_sample = n_friends, n_num = 50)
observed = gstat.convertNameNet2Observed(name_net)
prediction = convertObserved2Prediction(standards, observed)
pairs = unzipGroupPrediction(prediction)
# Use tester class
num_label = 4
tester = tests.tester(num_label)
assert(len(pairs) >= 2) # sanity check
tester.record(pairs[0], pairs[1])
tester.recordMeta(pairs[2])
#print "Accuracy:", tester.accuracy()
#print "Conf matrix:"
#print tester.confusionMatrix(friends.groupToNumeric)
return tester.accuracy()
def run(self):
standards = serial.objFromFile('prob_groups.txt')
c = ClusterBridge()
name_net = c.getNameNet()
observed = gstat.convertNameNet2Observed(name_net)
prediction = matchstat.convertObserved2Prediction(standards, observed)
# Format becomes:
# [(true1, true2,...), (predict1, predict2, ...), (meta1, meta2, ...)]
pairs = matchstat.unzipGroupPrediction(prediction)
num_label = 4
tester = tests.tester(num_label)
assert(len(pairs) >= 2) # sanity check
tester.record(pairs[0], pairs[1])
tester.recordMeta(pairs[2])
#print thres
print "Accuracy:", tester.accuracy()
print "Conf matrix:"
print tester.confusionMatrix(friends.groupToNumeric)
