Table 14. Test set results of M1–M4.

	Class	Precision	Recall	F1 score		Class	Precision	Recall	F1 score	Support
M1	1	0.8895	0.8635	0.8763	M3	1	0.8914	0.8652	0.8781	3,085
	2	0.9253	0.9239	0.9246		2	0.9259	0.9242	0.9250	9,956
	3	0.8681	0.9170	0.8919		3	0.8674	0.9189	0.8924	9,956
	4	0.8062	0.9170	0.8580		4	0.8236	0.8973	0.8589	9,956
	5	0.8993	0.6210	0.7347		5	0.8517	0.6658	0.7474	5,108
	6	0.4778	0.2170	0.2985		6	0.5079	0.2148	0.3019	447
	Accuracy			0.8671		Accuracy			0.8687	38,508
	Macro avg	0.8110	0.7432	0.7640		Macro avg	0.8113	0.7477	0.7673	38,508
M2	1	0.8820	0.8476		M4	1	0.8818	0.8804	0.8811	3,085
	2	0.9158	0.9226	0.9191		2	0.9293	0.9217	0.9255	9,956
	3	0.8601	0.9133	0.8859		3	0.8694	0.9166	0.8924	9,956
	4	0.8153	0.8896	0.8509		4	0.8216	0.9013	0.8596	9,956
	5	0.8492	0.6496	0.7361		5	0.8625	0.6656	0.7514	5,108
	6	0.5439	0.2081	0.3010		6	0.5026	0.2170	0.3031	447
	Accuracy			0.8611		Accuracy			0.8689	38,508
	Macro avg	0.8110	0.7385	0.7596		Macro avg	0.8112	0.7504	0.7688	38,508

Note:

Values in bold represent the optimum values for each group.