Using ordinal logistic regression to evaluate the performance of laser-Doppler predictions of burn-healing time

Table 7 Classification performance of the current palette (A) under allocation rules 1–2, (B), under rule 3, and (C) under rule 4, and (D) of the optimized palette under rule 1.

A) Rules 1 and 2	Pred. < 14 days	Pred. 14–21 days	Pred. > 21 days
Obs. < 14 days	162	24	0
Obs. 14–21 days	5	39	3
Obs > 21 days	0	3	25
	% wrong	% av. wrong per class
	13.4	13.6
B) Rule 3	Pred. < 14 days	Pred. 14–21 days	Pred. > 21 days
Obs. < 14 days	161	25	0
Obs. 14–21 days	5	39	3
Obs > 21 days	0	3	25
	% wrong	% av. wrong per class
	13.8	13.7
C) Rule 4	Pred. < 14 days	Pred. 14–21 days	Pred. > 21 days
Obs. < 14 days	177	9	0
Obs. 14–21 days	11	35	1
Obs > 21 days	0	7	21
	% wrong	% av. wrong per class
	10.7	18.5
D) Optimized	Pred. < 14 days	Pred. 14–21 days	Pred. > 21 days
Obs. < 14 days	178	8	0
Obs. 14–21 days	7	36	4
Obs > 21 days	0	2	26
	% wrong	% av. wrong per class
	8.1	11.6

ISSN: 1471-2288