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

Table 4

The performance comparison of the existing DNN model [9] and the proposed model.

		The existing DNN model [9]	The proposed model (SMOTE+ENN)
Accuracy	Training set	0.99	1.0
	5-fold cross-validation
	Test set	0.85	0.95
Precision	Training set	0.991	1.0
	5-fold cross-validation
	Test set	0.856	0.96
Recall	Training set	0.99	1.0
	5-fold cross-validation
	Test set	0.85	0.95
F1-score	Training set	0.99	1.0
	5-fold cross-validation
	Test set	0.85	0.952