Table 8. Evaluation results of software metrics-based models according to Viszkok et al.’s dataset (Viszkok, Hegedus & Ferenc, 2021) using both process and static metrics.

Approach	Precision (%)	Recall (%)	F1-score (%)
SDNN	94.2	69.3	80.8
CDNN	93.1	71.4	80.8
RF	96.7	72.8	83.1
KNN	97.0	63.9	77.0
Linear regression	91.0	28.5	43.5
Logistic regression	82.4	34.6	48.7
SVM	96.8	57.1	71.8
DT	90.9	75.3	82.4
NB	27.8	12.0	16.7
Our Stacking CNN classifiers	88.84	86.05	87.37
Our stacking CNN classifiers(RUS)	87.38	87.38	87.38
Our stacking CNN classifiers(ROS)	97.06	97.05	97.05

Notes.

The bold values are the best results among other classifiers.