Table 4.

Prediction performance of KNN, old versus new. "Old" = KNN results from [18], same as Table 2; "New" = KNN applied to the "relaxed datasets."

	Overall Error		False +		False -		DLR +		DLR -
Region	Old	New	Old	New	Old	New	Old	New	Old	New
1	0.34	0.36	0.34	0.35	0.36	0.41	1.88	1.69	0.55	0.63
2	0.32	0.33	0.31	0.29	0.38	0.46	2.01	1.85	0.55	0.65
3	0.31	0.34	0.27	0.30	0.50	0.48	1.84	1.70	0.68	0.70
4	0.32	0.32	0.30	0.24	0.35	0.47	2.15	2.21	0.50	0.62
5	0.23	0.32	0.18	0.30	0.53	0.45	2.57	1.84	0.65	0.64
6	0.28	0.31	0.24	0.27	0.41	0.42	2.40	2.11	0.55	0.58
7	0.30	0.32	0.28	0.26	0.37	0.46	2.29	2.12	0.51	0.61
8	0.31	0.35	0.28	0.33	0.42	0.42	2.08	1.75	0.58	0.63
Mean	0.30	0.33	0.28	0.29	0.42	0.45	2.15	1.91	0.57	0.63