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 pipelines(n_central, n_friends):
''' Includes all the pipelines and returns a pair of true label list and a
predicted label list. '''
# A list of standard distributions. Format is
# [(label, prob), (label, prob), ...]
#standards = serial.objFromFile('prob_groups.txt')
standards = gstat.generateEmpiricalDistributionFromSample(n_central, n_friends)
print standards
# Age and a set of names. Format is
# [(age, [names...]), (age, [names...])]
name_net = friends.getNameNetworkSampled(n_central, n_friends)
# Format becomes:
# [(age, prob, meta), (age, prob, meta), ...]
observed = gstat.convertNameNet2Observed(name_net)
# Format becomes:
# [(age, prediction, meta), ...]
prediction = convertObserved2Prediction(standards, observed)
# Format becomes:
# [(true1, true2,...), (predict1, predict2, ...), (meta1, meta2, ...)]
pairs = unzipGroupPrediction(prediction)
return pairs
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)
