melNET_average_conf_matrices_iter.py

# importing libs
import os
import csv
import numpy as np


#######################################################################################################################
# Main >>>
def main():
    start = input('Starting Iteration-Number: ')
    max_iter = input('Maximum Iteration-Number: ')
    freq = input('Iteration Frequency: ')

    for iter_num in range(int(start), int(max_iter) + int(freq), int(freq)):
        iter_root = os.getcwd() + '/iter_' + str(iter_num)
        avg_conf = open(iter_root + "/avg_conf.csv", 'w')
        avg_conf.write('Threshold, Specificity, FPR, TPR, AUC, Accuracy\n')

        specificity = np.zeros((1, 21))
        TPR = np.zeros((1, 21))
        accuracy = np.zeros((1, 21))

        for fold in range(5):
            temp_specificity = []
            temp_TPR = []
            temp_accuracy = []
            with open(iter_root + "/Fold_0" + str(fold+1) + "/conf_mat.csv") as conf:
                conf_reader = csv.reader(conf, delimiter=',')
                line_count = 0
                for row in conf_reader:
                    if line_count == 0:  # skipping 1st row
                        line_count += 1
                    else:
                        temp_specificity.append(float(row[6]))
                        temp_TPR.append(float(row[5]))
                        temp_accuracy.append(float(row[7]))

            np.reshape(temp_specificity, (1, -1))
            np.reshape(temp_TPR, (1, -1))
            np.reshape(temp_accuracy, (1, -1))

            specificity += temp_specificity
            TPR += temp_TPR
            accuracy += temp_accuracy

        specificity /= 5.0
        TPR /= 5.0
        accuracy /= 5.0

        np.reshape(specificity, (1, -1))
        np.reshape(TPR, (1, -1))
        np.reshape(accuracy, (1, -1))

        x = 0
        for thresh in range(0, 105, 5):
            avg_conf.write(str(thresh/100.0) + ',' + str(specificity[0][x]) + ',' + str(1-specificity[0][x]) +
                           ',' + str(TPR[0][x]) + ',' + 'AUC' + ',' + str(accuracy[0][x]) + '\n')
            x += 1

        avg_conf.close()


#######################################################################################################################
# Main Call Func. >>>
if __name__ == "__main__":
    main()