Deep Learning-Based Prediction of Physical Stability considering Class Imbalance for Amorphous Solid Dispersions

Table 2

The performance comparison of models applying various sampling techniques to the RF [9].

		Without sampling	Oversampling		Undersampling				Hybrid sampling
			SMOTE	ADASYN	TL	ENN	OSS	CNN	Smote+ENN	Smote+TL
Accuracy	Training set	1.0	0.985	0.973	1.0	1.0	1.0	1.0	1.0	0.987
	5-fold cross-validation
	Test set	0.85	0.9	0.85	0.8	0.65	0.6	0.5	0.9	0.9
Precision	Training set	1.0	0.985	0.975	1.0	1.0	1.0	1.0	1.0	0.987
	5-fold cross-validation
	Test set	0.83	0.933	0.83	0.77	0.842	0.739	0.647	0.933	0.933
Recall	Training set	1.0	0.985	0.973	1.0	1.0	1.0	1.0	1.0	0.987
	5-fold cross-validation
	Test set	0.85	0.9	0.85	0.8	0.65	0.6	0.5	0.9	0.9
F1-score	Training set	1.0	0.985	0.973	1.0	1.0	1.0	1.0	1.0	0.987
	5-fold cross-validation
	Test set	0.832	0.906	0.832	0.783	0.688	0.607	0.559	0.906	0.906