import pandas as pd
import numpy as np
import argparse as ap

cnn_class_0_probabilities = []
cnn_class_1_probabilities = []

gaussian_naive_bayes_class_0_probabilities = []
gaussian_naive_bayes_class_1_probabilities = []

cnn_predictions = 'test_predictions_cnn.csv'
gaussian_naive_bayes_predictions = 'test_predictions_gaussian_naive_bayes.csv'

cnn_predictions_csv_file = pd.read_csv(cnn_predictions)
gaussian_naive_bayes_predictions_csv_file = pd.read_csv(gaussian_naive_bayes_predictions)

df1 = pd.DataFrame(cnn_predictions_csv_file)
df2 = pd.DataFrame(gaussian_naive_bayes_predictions_csv_file)

for index, row in df1.iterrows():
	class_0_probability = row['0']
	class_1_probability = row['1']
	
	cnn_class_0_probabilities.append(class_0_probability)
	cnn_class_1_probabilities.append(class_1_probability)

for index, row in df2.iterrows():
	class_0_probability = row['0']
	class_1_probability = row['1']
	
	gaussian_naive_bayes_class_0_probabilities.append(class_0_probability)
	gaussian_naive_bayes_class_1_probabilities.append(class_1_probability)

total_prediction_probability = ((np.array(cnn_class_0_probabilities) * np.array(gaussian_naive_bayes_class_0_probabilities)) + (np.array(cnn_class_1_probabilities) * np.array(gaussian_naive_bayes_class_1_probabilities)))/2

#print total_prediction_probability

threshold = (np.max(total_prediction_probability) + np.mean(total_prediction_probability))/2

# decision

for prediction_probability in total_prediction_probability:
	if prediction_probability > threshold:
		print('irregular')
	elif prediction_probability < threshold:
		print('regular')