def matthews_correlation_coefficient(y_true, y_pred):
rrr, TP, FP, TN, FN, P, N = basics(y_true, y_pred,
matthews_correlation_coefficient)
if TP is not None:
rrr = mcc_basic(TP, FP, TN, FN)
elif rrr == _NON_BINARY:
y_true, y_pred = prep_ys(y_true, y_pred)
true_class_names = inv_map(nnstate.CURRENT_TRUE_MAP)
reduced_y_true = []
for y in y_true:
tru_nam = true_class_names[int(y)]
reduced_y_true.append(reduced_label(tru_nam))
reduced_y_pred = []
for y in y_pred:
tru_nam = true_class_names[int(y)]
reduced_y_pred.append(reduced_label(tru_nam))
TP, FP, TN, FN, P, N = binary_results(reduced_y_true, reduced_y_pred)
if TP is not None:
rrr = mcc_basic(TP, FP, TN, FN)
return update_met_log(matthews_correlation_coefficient, rrr, inc=True)
def sensitivity(y_true, y_pred):
rrr, TP, FP, TN, FN, P, N = basics(y_true, y_pred, sensitivity)
if TP is not None:
if P == 0:
rrr = _EMPTY_TENSOR
else:
rrr = TP / P
return update_met_log(sensitivity, rrr)
def miss_rate(y_true, y_pred):
rrr, TP, FP, TN, FN, P, N = basics(y_true, y_pred, miss_rate)
if TP is not None:
if P == 0:
rrr = _EMPTY_TENSOR
else:
rrr = FN / P
return update_met_log(miss_rate, rrr)
def specificity(y_true, y_pred):
rrr, TP, FP, TN, FN, P, N = basics(y_true, y_pred, specificity)
if TP is not None:
if N == 0:
rrr = -2
else:
rrr = TN / N
return update_met_log(specificity, rrr)
def false_positive_rate(y_true, y_pred):
rrr, TP, FP, TN, FN, P, N = basics(y_true, y_pred, false_positive_rate)
if TP is not None:
if N == 0:
rrr = _EMPTY_TENSOR
else:
rrr = FP / N
return update_met_log(false_positive_rate, rrr)
def precision(y_true, y_pred):
rrr, TP, FP, TN, FN, P, N = basics(y_true, y_pred, precision)
if TP is not None:
if (TP + FP) == 0:
rrr = _EMPTY_TENSOR
else:
rrr = TP / (TP + FP)
return update_met_log(precision, rrr)
def error_rate(y_true, y_pred, real_error=True):
rrr, TP, FP, TN, FN, P, N = basics(y_true, y_pred, error_rate)
# breakpoint()
if TP is not None:
if (P + N) == 0:
rrr = _EMPTY_TENSOR
else:
rrr = error_rate_basic(FP, FN, P, N)
elif rrr == _NON_BINARY:
y_true, y_pred = prep_ys(y_true, y_pred)
rrr = error_rate_core(y_true, y_pred)
return update_met_log(error_rate, rrr) if real_error else rrr
def mcc_multi(y_true, y_pred):
rrr, TP, FP, TN, FN, P, N = basics(y_true, y_pred, mcc_multi)
if TP is not None:
err('this should never happen if i have >2 classes')
rrr = mcc_basic(TP, FP, TN, FN)
elif rrr == _NON_BINARY:
y_true, y_pred = prep_ys(y_true, y_pred)
import sklearn.metrics
# https://github.com/scikit-learn/scikit-learn/issues/16924
go = True
for i in range(nnstate.num_pred_classes()):
if count_nonzero(y_pred == i) == 0:
go = False
break
if go:
rrr = sklearn.metrics.matthews_corrcoef(list(map(int, y_true)),
y_pred.tolist())
else:
rrr = -6
return update_met_log(mcc_multi, rrr, inc=True)
def accuracy(y_true, y_pred):
rrr = 1 - error_rate(y_true, y_pred, real_error=False)
return update_met_log(accuracy, rrr)
