Machine Learning Applied to Near-Infrared Spectra for Chicken Meat Classification

Table 2

Wavelengths selected by correlation-based feature selection (CFS), Decision Table (DTable), REPTree, and M5P algorithms using pH and compared by coefficient of correlation (CC), mean absolute error (MAE), root mean squared error (RMSE), and merit.

Algorithm		Wavelengths	CC	MAE	RMSE	Merit
CFS	pH	(608, 684)	—	—	—	0.794
		(444, 610)	—	—	—	0.939
DTable	pH	(600, 602, 604, 608, 622)	0.7008	0.0859	0.1124	—
		(442)	0.9120	1.8820	2.6369	—
REPTree	pH	(568, 606, 654, 666, 698)	0.7322	0.0833	0.1062	—
		(442, 452, 460, 604, 626, 1354, 1880)	0.9033	2.0914	2.7692	—
M5P	pH	(400, 448, 482, 540, 608, 626, 968, 1376, 1710, 1874, 2476, 2494)	0.7977	0.0702	0.0936	—
		(430, 410, 618, 1880)	0.9485	1.5990	2.0341	—