def make_submission():
print("Reading data")
fea, status = features.online_extract_features('data/train.csv',
limit=5e6)
print("Training the model")
rf = RandomForestClassifier(n_estimators=50,
verbose=2,
compute_importances=True,
oob_score=True,
#criterion='entropy',
n_jobs=1)
rf.fit(fea, status['OpenStatus'])
print "Features Importance:"
imps = zip(rf.feature_importances_,
fea.keys())
imps.sort(reverse=True)
print '\n'.join([ str(_) for _ in imps ])
print "Generalization Error:", rf.oob_score_
print("Reading test file and making predictions")
test_features = features.online_extract_features('data/'+test_file,
train=False,
limit=1e12)[0]
probs = rf.predict_proba(test_features)
if True:
print("Calculating priors and updating posteriors")
new_priors = cu.get_priors(full_train_file)
old_priors = cu.get_priors(train_file)
probs = cu.cap_and_update_priors(old_priors, probs, new_priors, 0.001)
print("Saving submission to %s" % submission_file)
cu.write_submission(submission_file, probs)
def make_submission():
print("Reading data")
fea, status = features.online_extract_features('data/train.csv',
train=True,
limit=1e9)
_dim(fea,'fea')
print("Training Level 1 : Open/Rest model")
open_status = [ or_binarize(e) for e in status['OpenStatus'] ]
is_not_open_status = [ s != 'open' for s in open_status ]
or_model = learn(fea,open_status)
print("Training Level 2 : Not Open Split model")
not_open_status = [ status['OpenStatus'][i] for i in range(len(is_not_open_status)) if is_not_open_status[i] ]
no_fea = fea[is_not_open_status]
_dim(no_fea,'no_fea')
no_model = learn(no_fea,not_open_status)
print("Reading test file and making predictions")
test_features = features.online_extract_features('data/'+test_file,
train=False,
limit=1e9)[0]
_dim(test_features,'test_features')
or_probs = or_model.predict_proba(test_features)
probs = []
for i in range(0,len(or_probs)):
or_prob = or_probs[i]
if or_prob[0] > or_prob[1]:
probs.append(np.array([1.0,0.0,0.0,0.0,0.0]))
else:
f = [ test_features[ff][i] for ff in test_features.keys() ]
a = no_model.predict_proba(f)
aa = np.insert(a,0,[0.0])
probs.append(aa)
probs = np.array(probs)
if False:
print("Calculating priors and updating posteriors")
new_priors = cu.get_priors(full_train_file)
old_priors = cu.get_priors(train_file)
probs = cu.cap_and_update_priors(old_priors, probs, new_priors, 0.001)
print("Saving submission to %s" % submission_file)
cu.write_submission(submission_file, probs)
