Table 2 Predicted errors of RF, GB, XGB,... | Emerald Publishing

Table 2

Predicted errors of RF, GB, XGB, CatBoost, Stacking and GAM

	Training/test dataset: 70–30%						Training/test dataset: 60–40%
	Base learners				Meta learner	Bench-mark	Base learners				Meta learner	Bench-mark
	RF	GB	XGB	CatBoost	Stacking	GAM	RF	GB	XGB	CatBoost	Stacking	GAM
Training MAE	0.70154	0.37991	0.41722	0.46168	0.43706	1.36310	0.70563	0.43931	0.45577	0.47990	0.45620	1.39216
Test MAE	1.24579	1.03017	1.07755	0.97881	0.96270	1.50149	1.25653	1.05379	1.08361	0.98311	0.97197	1.50892
Training MSE	0.92544	0.36585	0.30289	0.36945	0.34814	3.79013	0.93198	0.49329	0.35565	0.40144	0.37481	3.94446
Test MSE	3.63813	2.71538	3.03371	2.63263	2.54754	5.29204	3.72530	2.85965	3.20808	2.77902	2.64387	5.44161
Training R Squared	0.98597	0.99445	0.99541	0.99440	0.99472	0.94162	0.98584	0.99251	0.99460	0.99390	0.99431	0.94009
Test R Squared	0.94818	0.96165	0.95639	0.96215	0.96338	0.92392	0.94770	0.95889	0.95581	0.96076	0.96266	0.92315

	Training/test dataset: 70–30%						Training/test dataset: 60–40%
	Base learners				Meta learner	Bench-mark	Base learners				Meta learner	Bench-mark
	RF	GB	XGB	CatBoost	Stacking	GAM	RF	GB	XGB	CatBoost	Stacking	GAM
Training MAE	0.70154	0.37991	0.41722	0.46168	0.43706	1.36310	0.70563	0.43931	0.45577	0.47990	0.45620	1.39216
Test MAE	1.24579	1.03017	1.07755	0.97881	0.96270	1.50149	1.25653	1.05379	1.08361	0.98311	0.97197	1.50892
Training MSE	0.92544	0.36585	0.30289	0.36945	0.34814	3.79013	0.93198	0.49329	0.35565	0.40144	0.37481	3.94446
Test MSE	3.63813	2.71538	3.03371	2.63263	2.54754	5.29204	3.72530	2.85965	3.20808	2.77902	2.64387	5.44161
Training R Squared	0.98597	0.99445	0.99541	0.99440	0.99472	0.94162	0.98584	0.99251	0.99460	0.99390	0.99431	0.94009
Test R Squared	0.94818	0.96165	0.95639	0.96215	0.96338	0.92392	0.94770	0.95889	0.95581	0.96076	0.96266	0.92315

	Training/test dataset: 50–50%
	Base learners				Meta learner	Bench-mark
	RF	GB	XGB	CatBoost	Stacking	GAM
Training MAE	0.70576	0.46162	0.46868	0.51 344	0.46760	1.40881
Test MAE	1.26267	1.06490	1.08805	0.99500	0.97412	1.53278
Training MSE	0.93900	0.51414	0.37927	0.45142	0.39451	4.06723
Test MSE	3.86655	3.15064	3.29724	3.11162	2.96726	5.88229
Training R Squared	0.98554	0.99208	0.99416	0.99305	0.99392	0.93833
Test R Squared	0.94739	0.95778	0.95470	0.95830	0.96023	0.92117

	Training/test dataset: 50–50%
	Base learners				Meta learner	Bench-mark
	RF	GB	XGB	CatBoost	Stacking	GAM
Training MAE	0.70576	0.46162	0.46868	0.51 344	0.46760	1.40881
Test MAE	1.26267	1.06490	1.08805	0.99500	0.97412	1.53278
Training MSE	0.93900	0.51414	0.37927	0.45142	0.39451	4.06723
Test MSE	3.86655	3.15064	3.29724	3.11162	2.96726	5.88229
Training R Squared	0.98554	0.99208	0.99416	0.99305	0.99392	0.93833
Test R Squared	0.94739	0.95778	0.95470	0.95830	0.96023	0.92117

Source(s): Table by the author