def classify_new_data(x_new):
"""
Loads the pickled classifier and tfidf vectorizer then uses those for transforming and classifying new data
:param x_new: New data that needs to be classified
:return: prediction on new data
"""
file_path_dict = MyUtils.get_file_path_dict()
with open(file_path_dict[CONSTANTS.TRAINED_MODEL], 'rb') as file:
classifier = pickle.load(file)
with open(file_path_dict[CONSTANTS.TFIDF_VECTORIZER], 'rb') as file:
tfidf_vect = pickle.load(file)
predicted = classifier.predict(tfidf_vect.transform(x_new))
return predicted
def train_model(X_data, y_data):
"""
Trains the model by using the parameter values. Default classifier is MultinomialNB
Also stores the trained model and fitted tfidf vectorizer on disk with file names 'trained_model' and 'tfidf_vect' under utility dir
:return: Fitted model and fitted tfidf_vectorizer
"""
file_path_dict = MyUtils.get_file_path_dict()
tfidf_vect = get_tfidf_vectorizer()
classifier = get_classifier()
X_tfidfed = tfidf_vect.fit_transform(X_data)
classifier.fit(X_tfidfed, y_data)
with open(file_path_dict[CONSTANTS.TRAINED_MODEL], 'wb') as file:
pickle.dump(classifier, file)
with open(file_path_dict[CONSTANTS.TFIDF_VECTORIZER], 'wb') as file:
pickle.dump(tfidf_vect, file)
return classifier, tfidf_vect