def testRegression():
y = data.loadSVMLightTarget("../feature_set/synthetic.reg")
# we are using pseudo index instead of the real feature set to use the
# precomputed kernel.
x = range(y.shape[0])
dataset = zip(x, y)
random.shuffle(dataset)
# this precomputed kernel reads the index value and returns the covariance
# matrix
kernel = kernel_factory.precomputed("../feature_set/synthetic.kernel")
for train, test in data.kFolds(dataset):
x_train, y_train = zip(*train)
x_train = np.atleast_2d(np.array(x_train)).T
y_train = np.atleast_2d(np.array(y_train)).T
reg = regress.GPRegressor(kernel)
reg.fit(x_train, y_train)
print "fitted"
# test
x_test, y_test = zip(*test)
x_test = np.atleast_2d(np.array(x_test)).T
y_test = np.atleast_2d(np.array(y_test)).T
y_predict = reg.predict(x_test)
print x_test.T
print y_predict.T
break
def run(self):
#dataset = self.getFollowingData()
dataset = self.loadData()
for train, test in data.kFolds(dataset):
# training
m = self.trainModelOn(train)
predicates = self.getPredicates(train)
e = self.getMostEffective(m, predicates)
print e
def getSeeds(self):
seeds = []
dataset = self.loadData()
for train, test in data.kFolds(dataset):
# training
m = self.trainModelOn(train)
predicates = self.getPredicates(train)
e = self.getMostEffective(m, predicates)
for tuple_list in e.values():
user_ids = map(operator.itemgetter(0), tuple_list)
seeds.extend(user_ids)
return list(set(seeds))
def outputOnlyMatched():
maxent.set_verbose(1)
text_dataset = getTextData(False)
following_dataset = getFollowingData(False)
dataset = zip(text_dataset, following_dataset)
random.shuffle(dataset)
print 'finished loading dataset'
tester = tests.tester(4)
n_total = 0
n_emit = 0
for train, test in data.kFolds(dataset):
text_train, following_train = zip(*train)
# training
t_model = trainedModelOn(text_train)
f_model = trainedModelOn(following_train)
# prediction
trials = []
for datum in test:
text_datum, following_datum = datum
text_context, target, weight = text_datum
following_context, target, weight = following_datum
t_pre = t_model.predict(text_context)
f_pre = f_model.predict(following_context)
if t_pre == f_pre:
trials.append((target, t_pre))
n_emit += 1
n_total += 1
trials = zip(*trials)
tester.record(trials[0], trials[1])
print 'accuracy:', tester.accuracy()
print 'confusion matrix:'
print tester.confusionMatrix()
print 'emitted portion:', float(n_emit) / float(n_total)
def getFreqplot(self, dataset):
n_buckets = 20
cali_count = np.zeros((4, n_buckets))
for train, test in data.kFolds(dataset):
# training
m = self.trainedModelOn(train)
for datum in test:
context, target, weight = datum
pre = m.eval_all(context)
for label, score in pre:
cali_count[int(label), int(floor(n_buckets * score))] += 1
return cali_count
def scoreDistribution():
maxent.set_verbose(1)
text_dataset = getTextData()
for train, test in data.kFolds(text_dataset):
model = trainedModelOn(train)
for datum in test:
context, target, weight = datum
pred = model.predict(context)
model.eval_all(context)
if pred != target:
prob = map(itemgetter(1),
sorted(model.eval_all(context), key = itemgetter(0)))
print prob, target
break
def doCrossValidation(self, dataset, size_limit):
tester = tests.tester(4)
for train, test in data.kFolds(dataset):
# training
train = random.sample(train, size_limit)
m = self.trainedModelOn(train)
# prediction
trials = []
for datum in test:
context, target, weight = datum
pre_target = m.predict(context)
trials.append((target, pre_target))
trials = zip(*trials)
tester.record(trials[0], trials[1])
print size_limit, tester.accuracy()
def seePredictionOnTrainingData():
maxent.set_verbose(1)
dataset = getTextData()
print 'finished loading dataset'
for train, test in data.kFolds(dataset):
m = trainedModelOn(train)
print "Accuracy on Training Set"
for datum in train:
context, target, weight = datum
print m.eval_all(context)
print "Accuracy on Test Set"
for datum in test:
context, target, weight = datum
print m.eval_all(context)
break
def regressAgeWithGP():
y = data.loadSVMLightTarget("../feature_set/text.sreg")
# we are using pseudo index instead of the real feature set to use the
# precomputed kernel.
x = range(y.shape[0])
dataset = zip(x, y)
random.shuffle(dataset)
# this precomputed kernel reads the index value and returns the covariance
# matrix
for tau in [1e-3, 1e-2, 1e-1, 1, 1e1, 1e2, 1e3]:
kernel = kernel_factory.precomputed("../feature_set/text.kernel", tau)
rms_errors = []
diviations = []
for train, test in data.kFolds(dataset):
x_train, y_train = zip(*train)
x_train = np.atleast_2d(np.array(x_train)).T
y_train = np.atleast_2d(np.array(y_train)).T
reg = regress.GPRegressor(kernel)
reg.fit(x_train, y_train)
# test
x_test, y_test = zip(*test)
x_test = np.atleast_2d(np.array(x_test)).T
y_test = np.atleast_2d(np.array(y_test)).T
y_predict = reg.predict(x_test)
rms_e = sqrt(np.mean((y_predict - y_test) ** 2))
div_e = np.mean(np.absolute(y_predict - y_test))
rms_errors.append(rms_e)
diviations.append(div_e)
print "tau:", tau
print "rms error:", np.mean(np.array(rms_e))
print "diviation:", np.mean(np.array(div_e))
def doCrossValidation(dataset):
tester = tests.tester(4)
for train, test in data.kFolds(dataset):
# training
m = trainedModelOn(train)
print 'train size', len(train)
# prediction
trials = []
for datum in test:
context, target, weight = datum
pre_target = m.predict(context)
trials.append((target, pre_target))
trials = zip(*trials)
tester.record(trials[0], trials[1])
print 'accuracy:', tester.accuracy()
print 'confusion matrix:'
print tester.confusionMatrix()
def simpleEnsemble(pickup):
maxent.set_verbose(1)
text_dataset = getTextData(False)
following_dataset = getFollowingData(False)
dataset = zip(text_dataset, following_dataset)
random.shuffle(dataset)
print 'finished loading dataset'
tester = tests.tester(4)
for train, test in data.kFolds(dataset):
text_train, following_train = zip(*train)
# training
t_model = trainedModelOn(text_train)
f_model = trainedModelOn(following_train)
# prediction
trials = []
for datum in test:
text_datum, following_datum = datum
text_context, target, weight = text_datum
following_context, target, weight = following_datum
t_conf = t_model.eval_all(text_context)
f_conf = f_model.eval_all(following_context)
pre_target = str(pickup(t_conf, f_conf))
trials.append((target, pre_target))
trials = zip(*trials)
tester.record(trials[0], trials[1])
print 'accuracy:', tester.accuracy()
print 'confusion matrix:'
print tester.confusionMatrix()
def run(self):
dataset = self.loadData()
f = open('confused.txt', 'w')
for train, test in data.kFolds(dataset):
# training
m = self.trainedModelOn(train)
dbcon = mlm.DBConnector()
for datum in test:
context, target, weight, meta = datum
pre = m.eval_all(context)
label = pre[0][0]
confi = pre[0][1]
if .2 <= confi <= .4 and label != target:
f.write('###')
f.write(' '.join(
map(str, [target, label,
meta['user_id'], meta['screen_name'], confi])))
f.write('\n')
#for predicate, value in context:
# f.write(predicate + ' ' + str(value) + '\n')
user_id = meta['user_id']
text, length = dbcon.loadText(user_id)
f.write(text)
f.write('\n\n\n')
f.close()
def classifierEnsemble():
text_dataset = getTextData(False)
following_dataset = getFollowingData(False)
dataset = zip(text_dataset, following_dataset)
random.shuffle(dataset)
print 'finished loading dataset'
tester = tests.tester(4)
def _conf_to_feature(conf1, conf2):
def _append_to_key(c):
def _append(f):
return (c + f[0], f[1])
return _append
conf1 = map(_append_to_key('0'), conf1)
conf2 = map(_append_to_key('1'), conf2)
confs = conf1
confs.extend(conf2)
return confs
for train, test in data.kFolds(dataset):
coffset = int(len(train) * .8)
text_train, following_train = zip(*train[:coffset])
# training
t_model = trainedModelOn(text_train)
f_model = trainedModelOn(following_train)
# train a chooser
chooser = cmaxent.MaxentModel()
chooser.begin_add_event()
for datum in train[coffset:]:
text_datum, following_datum = datum
text_context, target, weight = text_datum
following_context, target, weight = following_datum
t_conf = t_model.eval_all(text_context)
f_conf = f_model.eval_all(following_context)
confs = _conf_to_feature(t_conf, f_conf)
chooser.add_event(confs, target)
chooser.end_add_event(0)
chooser.train(50, 'lbfgs', 1e-1, 1e-4)
# retrain the underlying classifiers
text_train, following_train = zip(*train)
t_model = trainedModelOn(text_train)
f_model = trainedModelOn(following_train)
# prediction
trials = []
for datum in test:
text_datum, following_datum = datum
text_context, target, weight = text_datum
following_context, target, weight = following_datum
t_conf = t_model.eval_all(text_context)
f_conf = f_model.eval_all(following_context)
confs = _conf_to_feature(t_conf, f_conf)
pre_target = chooser.predict(confs)
trials.append((target, pre_target))
trials = zip(*trials)
tester.record(trials[0], trials[1])
print 'accuracy:', tester.accuracy()
print 'confusion matrix:'
print tester.confusionMatrix()