Table 3.

Performance of the XGBoost models for individual sheets based on fivefold cross-validation.

Sheet	Mesoregion	Number of landforms	Accuracy	Kappa coefficient
Świnoujście	Szczecin Coastland	19	0.961	0.957
Toruń	Chełmno–Dobrzyń Lakeland, Toruń–Eberswalde Ice Marginal Valley	25	0.913	0.908
Kutno	Central Masovia Lowland, Southern Wielkopolska Lowland	15	0.958	0.952
Jelenia Góra	Western Sudety, Mountains	19	0.962	0.958
Tomaszów Lubelski	Roztocze Upland, Sandomierz Basin	15	0.880	0.866
Katowice	Silesia Upland, Woźniki–Wieluń Upland	13	0.943	0.936
Kraków	Krakow–Częstochowa Upland, Kraków Gate	17	0.949	0.945
Nowy Targ	Orawa–Podhale Basin, Tatra Range	16	0.940	0.934

The physico-geographical mesoregions are defined based on Solon et al.⁵⁶ classification.