def load(self, path):
classes = load_pickle(path, "steps.pickle")
self._steps = [object.__new__(c) for c in classes]
for index, step in enumerate(self._steps):
step_path = path + str(index)
step.load(step_path)
self._fitted_scores = np.sort(load_numpy_txt(path+"scores"))
def transform(self, x=None):
if not x:
data = load_numpy_txt(self._test_input_path)
y_true = data[:, 0]
y_score = data[:, 1]
y_pred = data[:, 2]
# Accuracy classification score.
accuracy = sklearn.metrics.accuracy_score(y_true, y_pred)
# Compute average precision (AP) from prediction scores
avg_precision = sklearn.metrics.average_precision_score(y_true,
y_score)
# Compute the F1 score, also known as balanced F-score or F-measure
f1 = sklearn.metrics.f1_score(y_true,
y_pred)
# fbeta = sklearn.metrics.fbeta_score(y_true, y_pred, beta=0.5) # Compute the F-beta score
# Compute the Matthews correlation coefficient (MCC) for binary classes
mcc = sklearn.metrics.matthews_corrcoef(y_true,
y_pred)
precision = sklearn.metrics.precision_score(y_true, y_pred) # Compute the precision
recall = sklearn.metrics.recall_score(y_true, y_pred) # Compute the recall
# Compute Area Under the Curve (AUC) from prediction scores
auc = sklearn.metrics.roc_auc_score(y_true,
y_score)
res = {
'accuracy': accuracy,
'avg_precision': avg_precision,
'f1': f1,
'mcc': mcc,
'precision': precision,
'recall': recall,
'auc': auc
}
save_json(res, self._test_output_path)
else:
raise NotImplemented