def bayes_generate_model(args):
'''
Function creates model for bayesian classifier
'''
bt = BayesianTest(dbfile=args.db_file, max_token_size=args.max_token_size)
if args.feats is not None:
features = eval(args.feats)
if isinstance(features,dict):
e = Entry(id=None, guid=None, entry=None, language=None)
if not e.check_feats(features):
print 'Incorrect format of feature dictionary'
return
else:
features = bt.get_best_features(count=args.count, n_fold_cv=args.n_fold_cv)
bt.create_model(args.model, used_features=features, count=args.count)
def bayes_test(args):
'''
Function starts test of bayesian classifier with given dataset and classifier
parameters.
'''
bt = BayesianTest(dbfile=args.db_file, max_token_size=args.max_token_size)
if args.feats is not None:
features = eval(args.feats)
if isinstance(features,dict):
e = Entry(id=None, guid=None, entry=None, language=None)
if not e.check_feats(features):
print 'Incorrect format of feature dictionary'
return
else:
features = bt.get_best_features(count=args.count, n_fold_cv=args.n_fold_cv)
bt.run(features=features, count=args.count, n_fold_cv=args.n_fold_cv)
def svm_classify(args):
'''
Manually classify given text with some SVM model
'''
# load model and create classifier
svm = SVM(kernel=None, C=None)
token_list = svm.load_model(args.model)
# convert text to vector X
entry = Entry(id=None, guid=None, entry=args.text, language=None,
max_token_size=1)
X = np.zeros((1,len(token_list)))
for token in entry.get_token_all():
if token.get_data_str() in token_list:
X[0][token_list.index(token.get_data_str())] = 1
# classify text
print int(svm.predict(X)[0])