Heritage Science

Table 1 Training, validation, and cross-validation matrices for the best performing LDA models for identification of lignin, using 5 sheets (above) and 1 sheet (below)

From: Characterisation of 19th and 20th century Chinese paper

LDA model	Training					Validation				Cross-validation
LDA model	Low (0/1)	High (2)	Total	% correct	Pre-processing	Low (0/1)	High (2)	Total	% correct	Low (0/1)	High (2)	Total	% correct
Low (0/1)	71	9	80	89	f’(χ), SNV, 4 PC scores	43	0	43	100	71	9	80	89
High (2)	4	24	28	86		3	8	11	73	4	24	28	86
Total	75	33	108	88		46	8	54	94	75	33	108	88
Low (0/1)	72	8	80	90	f’(χ), MSC, 5 PC scores	42	1	43	98	71	9	80	89
High (2)	4	24	28	86		1	10	11	91	5	23	28	82
Total	76	32	108	89		43	11	54	96	76	32	108	87

When calculating first derivative (f’(χ)), the window width of 5 was used

Back to article page