def score(self, cutoff: float = 0.5, method: str = "accuracy"):
if (method in ("accuracy", "acc")):
return (accuracy_score(self.y, self.deploySQL(cutoff), self.test_relation, self.cursor))
elif (method == "auc"):
return auc(self.y, self.deploySQL(), self.test_relation, self.cursor)
elif (method == "prc_auc"):
return prc_auc(self.y, self.deploySQL(), self.test_relation, self.cursor)
elif (method in ("best_cutoff", "best_threshold")):
return (roc_curve(self.y, self.deploySQL(), self.test_relation, self.cursor, best_threshold = True))
elif (method in ("recall", "tpr")):
return (recall_score(self.y, self.deploySQL(cutoff), self.test_relation, self.cursor))
elif (method in ("precision", "ppv")):
return (precision_score(self.y, self.deploySQL(cutoff), self.test_relation, self.cursor))
elif (method in ("specificity", "tnr")):
return (specificity_score(self.y, self.deploySQL(cutoff), self.test_relation, self.cursor))
elif (method in ("negative_predictive_value", "npv")):
return (precision_score(self.y, self.deploySQL(cutoff), self.test_relation, self.cursor))
elif (method in ("log_loss", "logloss")):
return (log_loss(self.y, self.deploySQL(), self.test_relation, self.cursor))
elif (method == "f1"):
return (f1_score(self.y, self.deploySQL(cutoff), self.test_relation, self.cursor))
elif (method == "mcc"):
return (matthews_corrcoef(self.y, self.deploySQL(cutoff), self.test_relation, self.cursor))
elif (method in ("bm", "informedness")):
return (informedness(self.y, self.deploySQL(cutoff), self.test_relation, self.cursor))
elif (method in ("mk", "markedness")):
return (markedness(self.y, self.deploySQL(cutoff), self.test_relation, self.cursor))
elif (method in ("csi", "critical_success_index")):
return (critical_success_index(self.y, self.deploySQL(cutoff), self.test_relation, self.cursor))
else:
raise ValueError("The parameter 'method' must be in accuracy|auc|prc_auc|best_cutoff|recall|precision|log_loss|negative_predictive_value|specificity|mcc|informedness|markedness|critical_success_index")
def score(self,
pos_label=None,
cutoff: float = 0.5,
method: str = "accuracy"):
pos_label = self.classes[1] if (
pos_label == None and len(self.classes) == 2) else pos_label
input_relation = self.deploySQL() + " WHERE predict_knc = '{}'".format(
pos_label)
y_score = "(CASE WHEN proba_predict > {} THEN 1 ELSE 0 END)".format(
cutoff)
y_proba = "proba_predict"
y_true = "DECODE({}, '{}', 1, 0)".format(self.y, pos_label)
if (pos_label not in self.classes):
raise ValueError(
"'pos_label' must be one of the response column classes")
elif (cutoff >= 1 or cutoff <= 0):
raise ValueError("'cutoff' must be in ]0;1[")
if (method in ("accuracy", "acc")):
return (accuracy_score(y_true, y_score, input_relation,
self.cursor))
elif (method == "auc"):
return (auc(y_true, y_proba, input_relation, self.cursor))
elif (method == "prc_auc"):
return (prc_auc(y_true, y_proba, input_relation, self.cursor))
elif (method in ("best_cutoff", "best_threshold")):
return (roc_curve(y_true,
y_proba,
input_relation,
self.cursor,
best_threshold=True))
elif (method in ("recall", "tpr")):
return (recall_score(y_true, y_score, input_relation, self.cursor))
elif (method in ("precision", "ppv")):
return (precision_score(y_true, y_score, input_relation,
self.cursor))
elif (method in ("specificity", "tnr")):
return (specificity_score(y_true, y_score, input_relation,
self.cursor))
elif (method in ("negative_predictive_value", "npv")):
return (precision_score(y_true, y_score, input_relation,
self.cursor))
elif (method in ("log_loss", "logloss")):
return (log_loss(y_true, y_proba, input_relation, self.cursor))
elif (method == "f1"):
return (f1_score(y_true, y_score, input_relation, self.cursor))
elif (method == "mcc"):
return (matthews_corrcoef(y_true, y_score, input_relation,
self.cursor))
elif (method in ("bm", "informedness")):
return (informedness(y_true, y_score, input_relation, self.cursor))
elif (method in ("mk", "markedness")):
return (markedness(y_true, y_score, input_relation, self.cursor))
elif (method in ("csi", "critical_success_index")):
return (critical_success_index(y_true, y_score, input_relation,
self.cursor))
else:
raise ValueError(
"The parameter 'method' must be in accuracy|auc|prc_auc|best_cutoff|recall|precision|log_loss|negative_predictive_value|specificity|mcc|informedness|markedness|critical_success_index"
)