Credit-Risk Prediction Model Using Hybrid Deep—Machine-Learning Based Algorithms

Table 4

Performance of CNN and hybrid CNN—SVM/RF/DT (%).

Performance metrics	CNN	CNN–SVM	CNN–DT	CNN–RF
Accuracy	86.70	98.60	96.90	95.50
Precision	91.74	97.49	95.57	97.56
Recall	82.75	99.57	94.52	95.65
F1-score	87.01	98.52	96.59	95.04
AUC	87.72	98.55	95.51	96.96

Learning scheme 80 : 20, support vector machine with radial basis function and , random forest and decision tree with criterion entropy, max-depth = 5, and n-estimators = 100.