Systematic Framework to Predict Early-Stage Liver Carcinoma Using Hybrid of Feature Selection Techniques and Regression Techniques

Table 13

Random Forest-based Regression algorithms with accuracy and error rate during training and testing.

Models	Accuracy (R2-Score) training (↑)	Accuracy (R2-Score) testing (↑)	MSE training (↓)	MSE testing (↓)
Simple Random Forest Regression	0.8921	0.4851	0.0619	0.1412
Random Forest Regression with Forward Selection (Wrapper Method)	0.8923	0.5031	0.0618	0.1387
Random Forest Regression with Backward Elimination (Wrapper Method)	0.8923	0.5031	0.0618	0.1387
Random Forest Regression with Bidirectional Elimination (Wrapper Method)	0.8923	0.5031	0.0618	0.1387
Random Forest Regression with L1 Regularization (Embedded Method)	0.8903	0.4844	0.0624	0.1413
Random Forest Regression with L2 Regularization (Embedded Method)	0.8903	0.4844	0.0624	0.1413
Random Forest Regression with Pearson Correlation (Filter Method)	0.8917	0.4798	0.062	0.1419
Random Forest Regression with Constant Feature Elimination (Filter Method)	0.8912	0.4864	0.0622	0.141
Random Forest Regression with Quasi-Constant (Filter Method)	0.8907	0.4903	0.0623	0.1405
Random Forest Regression with Correlated Features (Filter Method)	0.8912	0.4864	0.0622	0.141
Random Forest Regression with Duplicate Features (Filter Method)	0.8912	0.4864	0.0622	0.141