Table 4 Core model performance under... | Emerald Publishing

Table 4

Core model performance under temporal and random validation

Validation	Model	Training N	Test N	MAE	RMSE	R-squared
Temporal holdout	Extra Trees	1,322	389	8.496	14.149	0.651
Temporal holdout	Random Forest	1,322	389	10.171	18.110	0.429
Temporal holdout	Gradient Boosting	1,322	389	10.809	18.932	0.376
Temporal holdout	Ridge	1,322	389	14.214	19.868	0.312
Temporal holdout	Mean benchmark	1,322	389	15.736	24.299	−0.029
Random observation split	Mean benchmark	1,368	343	14.939	21.859	−0.001
Random observation split	Ridge	1,368	343	10.385	16.635	0.420
Random observation split	Random Forest	1,368	343	3.874	5.701	0.932
Random observation split	Extra Trees	1,368	343	3.197	4.938	0.949
Random observation split	Gradient Boosting	1,368	343	8.238	15.569	0.492

Validation	Model	Training N	Test N	MAE	RMSE	R-squared
Temporal holdout	Extra Trees	1,322	389	8.496	14.149	0.651
Temporal holdout	Random Forest	1,322	389	10.171	18.110	0.429
Temporal holdout	Gradient Boosting	1,322	389	10.809	18.932	0.376
Temporal holdout	Ridge	1,322	389	14.214	19.868	0.312
Temporal holdout	Mean benchmark	1,322	389	15.736	24.299	−0.029
Random observation split	Mean benchmark	1,368	343	14.939	21.859	−0.001
Random observation split	Ridge	1,368	343	10.385	16.635	0.420
Random observation split	Random Forest	1,368	343	3.874	5.701	0.932
Random observation split	Extra Trees	1,368	343	3.197	4.938	0.949
Random observation split	Gradient Boosting	1,368	343	8.238	15.569	0.492

Source(s): Computations from the study dataset (2026)