evaluation_metrics.py

import numpy as np
import copy
from math import sqrt
from scipy import stats
from sklearn import preprocessing, metrics
from sklearn.metrics import confusion_matrix

def prec_rec_f1_acc_mcc(y_true, y_pred):
    performance_threshold_dict  = dict()

    ### ADDED on 28 July 2020 by YIP YEW MUN ###
    y_true_tmp = []
    for each_y_true in y_true:
        y_true_tmp.append(each_y_true.item())
    y_true = y_true_tmp

    y_pred_tmp = []
    for each_y_pred in y_pred:
        y_pred_tmp.append(each_y_pred.item())
    y_pred = y_pred_tmp
    ### ADDED on 28 July 2020 by YIP YEW MUN ###

    precision = metrics.precision_score(y_true, y_pred)
    recall = metrics.recall_score(y_true, y_pred)
    f1_score = metrics.f1_score(y_true, y_pred)
    accuracy = metrics.accuracy_score(y_true, y_pred)
    mcc = metrics.matthews_corrcoef(y_true, y_pred)
    tn, fp, fn, tp = confusion_matrix(y_true, y_pred).ravel()
    performance_threshold_dict["Precision"] = precision
    performance_threshold_dict["Recall"] = recall
    performance_threshold_dict["F1-Score"] = f1_score
    performance_threshold_dict["Accuracy"] = accuracy
    performance_threshold_dict["MCC"] = mcc
    performance_threshold_dict["TP"] = tp
    performance_threshold_dict["FP"] = fp
    performance_threshold_dict["TN"] = tn
    performance_threshold_dict["FN"] = fn


    return performance_threshold_dict

def get_list_of_scores():
    return ["Precision", "Recall", "F1-Score", "Accuracy", "MCC", "TP", "FP", "TN", "FN"]