def test(data):
y = data["fraudulent"]
X = data.drop(['fraudulent'], axis=1)
split_point = int(0.8 * len(y))
X_train = X.iloc[:split_point]
X_test = X.iloc[split_point:]
y_train = y.iloc[:split_point]
y_test = y.iloc[split_point:]
clf = my_model()
clf.fit(X_train, y_train)
predictions = clf.predict(X_test)
eval = my_evaluation(predictions, y_test)
f1 = eval.f1(target=1)
return f1
def obj_func2(predictions, actuals, pred_proba=None):
# One objectives: higher f1 score
eval = my_evaluation(predictions, actuals, pred_proba)
return [eval.f1()]
def obj_func1(predictions, actuals, pred_proba=None):
# Two objectives: higher recall and higher precision
eval = my_evaluation(predictions, actuals, pred_proba)
return [eval.recall(), eval.precision()]
def obj_func2(predictions, actuals, pred_proba=None):
# Two objectives: higher recall and lower false positive rate
eval = my_evaluation(predictions, actuals, pred_proba)
return eval.f1()