Forecasting the Direction of Short-Term Crude Oil Price Changes with Genetic-Fuzzy Information Distribution

Table 5

The forecasting result of logistic regression.

Series	m	D	TN	TP	AUC	Optimal threshold
Brent	4	0.7000	0.8889	0.5455	0.6364	0.5584
Brent	5	0.7000	0.8889	0.5455	0.6566	0.5557
Brent	6	0.6500	0.7778	0.5455	0.6061	0.5574
WTI	4	0.6000	1.0000	0.1111	0.4444	0.6748
WTI	5	0.7000	1.0000	0.3333	0.5253	0.6436
WTI	6	0.7000	0.7273	0.6667	0.6465	0.5958

The bold estimators are optimal. is the directional accuracy, TN is true negative ratio, and TP is true positive ratio. The ROC (receiver operating characteristic) curve plots the true positive rate as a function of false positive rate for differing classification thresholds. AUC is used to measure the prediction performance that is the area under the ROC curve.