def N_Abcd(predicted, actual):
predicted_txt = []
# abcd = Abcd(db='Traing', rx='Testing')
global The
def isDef(x):
return "Defective" if x > 0 else "Non-Defective"
# use the.option.threshold for cart,
# rf and where!!
for data in predicted:
predicted_txt += [
isDef(data)] # this is for defect prediction, binary classes
# predicted_txt.append(data) # for multiple classes, just use it
score = sk_abcd(predicted_txt, actual)
return score
def _Abcd(predicted, actual):
predicted_txt = []
# abcd = Abcd(db='Traing', rx='Testing')
global The
def isDef(x):
return "Defective" if x >= The.option.threshold else "Non-Defective"
# use the.option.threshold for cart,
# rf and where!!
for data in predicted:
# predicted_txt += [isDef(data)] # this is for defect prediction, binary classes
predicted_txt.append(data) # for multiple classes, just use it
score = sk_abcd(predicted_txt, actual)
# if The.option.tunedobjective == 6: # auc
actual_binary = np.array([ 1 if i == "Delay" else 0 for i in actual ])
predicted_binary = np.array([ 1 if i == "Delay" else 0 for i in predicted ])
score[0].append(int(roc_auc_score(actual_binary,predicted_binary)*100))
score[1].append(int(roc_auc_score(actual_binary,predicted_binary)*100))
return score
def callModel(self, clf):
predict_result = clf.predict(self.test_X)
predict_pro = clf.predict_proba(self.test_X)
scores = sk_abcd(predict_result, self.test_Y, predict_pro[:, 1])
return scores[-1]
def callModel(self, clf, threshold):
predict_result = clf.predict(self.test_X)
# predict_pro = clf.predict_proba(self.test_X)
scores = sk_abcd(predict_result, self.test_Y, threshold)
return scores[-1]
