import os
from unittest import TestCase
import pandas
from sklearn.model_selection import cross_val_score
from sklearn.model_selection import train_test_split
from qas.classifier.question_classifier import classify_question
from qas.constants import CORPUS_DIR
from qas.corpus.data import QUESTION_CLASSIFICATION_TRAINING_DATA
class TestClassifyQuestion(TestCase):
classification_score = 0.60
def test_classify_question(self):
training_data_path = os.path.join(CORPUS_DIR, QUESTION_CLASSIFICATION_TRAINING_DATA)
df_question = pandas.read_csv(training_data_path, sep='|', header=0)
df_question_train, df_question_test = train_test_split(df_question, test_size=0.2, random_state=42)
predicted_class, clf, df_question_train_label, df_question_train = \
classify_question(df_question_train=df_question_train, df_question_test=df_question_test)
scores = cross_val_score(clf, df_question_train, df_question_train_label)
print("Accuracy: %0.2f (+/- %0.2f)" % (scores.mean(), scores.std() * 2))
print("SD:", scores.std())
assert scores.mean() > self.classification_score
