A Machine Learning-Based Model for Predicting Atmospheric Corrosion Rate of Carbon Steel

Table 4

Statistical quantifications of different predicted models.

Model	(mg/m^2.day)				(%)
	All	Train	Test	Val	All	Train	Test	Val	All	Train	Test	Val	All	Train	Test	Val
Knotkova et al. [19], Roberge, Klassen [21]	1.6201	1.6034	1.6835	1.6328	0.0281	0.0396	0.1947	0.1893	0.9950	0.9951	0.9946	0.9949	0.9095	0.9062	0.9371	0.8939
ISOCORRAG [25]	1.6316	1.6145	1.7080	1.6328	0.0283	0.0400	0.1986	0.1892	0.9949	0.9950	0.9944	0.9949	0.9048	0.9010	0.9284	0.8966
MICAT [25]	1.5287	1.5116	1.5548	1.5807	0.0263	0.0371	0.1783	0.1825	0.9955	0.9956	0.9954	0.9952	0.9112	0.9117	0.9250	0.8946
MLR1	1.6685	1.6653	1.6642	1.6879	0.0309	0.0440	0.2054	0.2082	0.9938	0.9939	0.9939	0.9937	0.8650	0.8621	0.8893	0.8526
MLR2	1.1730	1.1760	1.1831	1.1485	0.0208	0.0299	0.1402	0.1358	0.9974	0.9974	0.9973	0.9975	0.9661	0.9652	0.9740	0.9643
MLR3	1.1726	1.1760	1.1782	1.1508	0.0208	0.0299	0.1399	0.1362	0.9974	0.9974	0.9974	0.9975	0.9670	0.9657	0.9757	0.9661
ANN	0.2812	0.2824	0.2458	0.3072	0.0001	0.0002	0.0011	0.0023	0.9998	0.9998	0.9999	0.9998	0.9651	0.9638	0.9780	0.9585