def ovr_model(X_train, X_test, y_train, y_test, labels, seed, params=None):
# set parameters for pipeline for each estimator of the OvR model
if params is not None:
pipeline = Pipeline([
('count_vectorizer', CountVectorizer()),
('tf', TfidfTransformer()),
('svc', LinearSVC(dual=False, max_iter=10e4, random_state=seed)),
])
ovr = OneVsRestClassifier(pipeline)
ovr.estimators_ = [copy.deepcopy(pipeline) for i in range(len(labels))]
for idx in range(len(labels)):
ovr.estimators_[idx].set_params(**params[labels[idx]])
# use simpler model where each estimator of the OvR model is equal
else:
pipeline = Pipeline([
('count_vectorizer', CountVectorizer()),
('tf', TfidfTransformer(sublinear_tf=True)),
('svc',
LinearSVC(C=0.1,
penalty='l1',
class_weight='balanced',
dual=False,
max_iter=10e4,
random_state=seed)),
])
ovr = OneVsRestClassifier(pipeline)
# train and predict model
start_time = time.time()
ovr.fit(X_train, y_train)
prediction = ovr.predict(X_test)
stop_time = time.time()
# calculate scores
f1 = f1_score(y_test, prediction, average=None)
accuracy = jaccard_score(y_test, prediction, average=None)
return f1, accuracy, stop_time - start_time
