Research Article

Hybridization of Machine Learning Algorithms and an Empirical Regression Model for Predicting Debris-Flow-Endangered Areas

Table 3

Performance metrics of all the predictive models under fivefold cross-validation.

ModelTraining dataTesting data
R2RMSEMAER2RMSEMAE

CR1NLRM0.8040.0860.0570.5750.0570.047
MARS0.7830.0620.0450.4570.0700.055
RF0.9160.0440.0310.5560.0600.050
SVM0.6690.0790.0510.5730.0590.049
MARS–NLRM0.8180.0520.0340.7150.0480.036
RF–NLRM0.8710.0440.0290.7170.0440.030
SVM–NLRM0.7740.0580.0310.7320.0420.031

CR2NLRM0.7700.0830.0560.6380.0760.054
MARS0.6960.0710.0520.7470.0610.048
RF0.9130.0440.0320.7220.0660.049
SVM0.6610.0770.0500.6190.0760.056
MARS–NLRM0.7490.0570.0360.6880.0560.035
RF–NLRM0.8590.0430.0280.7270.0510.035
SVM–NLRM0.7460.0570.0360.7020.0540.033

CR3NLRM0.7680.0780.0530.2920.1040.069
MARS0.7720.0620.0490.1150.1170.084
RF0.9200.0400.0290.2520.1010.073
SVM0.6330.0800.0560.2330.1100.072
MARS–NLRM0.8500.0500.0330.5420.0740.045
RF–NLRM0.8990.0410.0280.5430.0740.047
SVM–NLRM0.8400.0520.0310.5400.0780.041

CR4NLRM0.8010.0810.0520.5530.0920.065
MARS0.6700.0730.0550.4760.1030.070
RF0.9130.0440.0320.5900.0820.061
SVM0.6210.0800.0550.3710.1270.081
MARS–NLRM0.7850.0510.0330.7190.0680.044
RF–NLRM0.8570.0420.0270.7440.0610.041
SVM–NLRM0.7380.0560.0320.7150.0650.039

CR5NLRM0.6600.0650.0560.8400.0760.051
MARS0.5830.0720.0540.5170.2570.150
RF0.8860.0410.0300.5970.1150.069
SVM0.5700.0750.0490.4940.1250.072
MARS–NLRM0.7130.0550.0340.8610.0600.035
RF–NLRM0.8310.0430.0290.7630.0810.043
SVM–NLRM0.7000.0570.0320.7690.0780.038