Potential of machine learning approaches for predicting mechanical properties of spruce wood in the transverse direction

Journal of Wood Science

Official Journal of the Japan Wood Research Society

Table 1 Prediction accuracy of conventional machine learning approaches and ANN using two splitting methods

Learning algorithm	Splitting method	Group	MOE (MPa)		MOR (MPa)
Learning algorithm	Splitting method	Group	R²	RMSE	R²	RMSE
kNN	Method #1	Train	0.879 (0.021)	71 (5)	0.870 (0.014)	0.47 (0.02)
	Method #1	Test	0.839 (0.065)	80 (12)	0.846 (0.044)	0.50 (0.04)
	Method #2	Train_A2,B1,B2	0.879	74	0.830	0.49
		Test_A1	− 0.657	176	− 1.532	1.30
		Train_A1,B1,B2	0.892	74	0.887	0.48
		Test_A2	0.534	82	− 0.347	0.84
		Train_A1,A2,B2	0.902	66	0.891	0.45
		Test_B1	0.847	72	0.617	0.57
		Train_A1,A2,B1	0.893	52	0.872	0.39
		Test_B2	0.701	141	0.498	0.92
SVM	Method #1	Train	0.870 (0.020)	74 (5)	0.857 (0.018)	0.49 (0.03)
	Method #1	Test	0.855 (0.051)	77 (8)	0.843 (0.063)	0.51 (0.08)
	Method #2	Train_A2,B1,B2	0.867	78	0.824	0.50
		Test_A1	0.406	106	− 0.430	0.98
		Train_A1,B1,B2	0.879	78	0.877	0.50
		Test_A2	0.710	65	0.028	0.71
		Train_A1,A2,B2	0.888	71	0.871	0.49
		Test_B1	0.763	89	0.635	0.55
		Train_A1,A2,B1	0.816	68	0.869	0.40
		Test_B2	0.703	141	0.555	0.86
RF	Method #1	Train	0.980 (0.003)	29 (1)	0.977 (0.003)	0.20 (0.01)
	Method #1	Test	0.862 (0.033)	75 (7)	0.806 (0.039)	0.57 (0.06)
	Method #2	Train_A2,B1,B2	0.979	31	0.971	0.20
		Test_A1	0.268	117	− 0.699	1.06
		Train_A1,B1,B2	0.979	33	0.979	0.21
		Test_A2	0.643	72	− 0.113	0.76
		Train_A1,A2,B2	0.985	26	0.980	0.19
		Test_B1	0.537	125	0.089	0.87
		Train_A1,A2,B1	0.986	19	0.980	0.16
		Test_B2	0.743	131	0.541	0.88
ANN	Method #1	Train	0.889 (0.019)	68 (5)	0.863 (0.014)	0.48 (0.02)
	Method #1	Test	0.868 (0.064)	72 (15)	0.841 (0.062)	0.51 (0.08)
	Method #2	Train_A2,B1,B2	0.884	72	0.815	0.51
		Test_A1	0.194	123	− 0.404	0.97
		Train_A1,B1,B2	0.882	77	0.850	0.55
		Test_A2	0.476	87	0.284	0.61
		Train_A1,A2,B2	0.890	71	0.836	0.55
		Test_B1	0.761	90	0.150	0.84
		Train_A1,A2,B1	0.842	63	0.805	0.48
		Test_B2	0.786	120	0.638	0.78