|
Actual class
|
TICA (p = 0.2)
|
TICA (p = 0.3)
|
NAUTICA
|
|---|
| |
NINT
|
INT
|
Recall
|
NINT
|
INT
|
Recall
|
NINT
|
INT
|
Recall
|
|---|
|
NINT
|
18
|
33
|
35%
|
18
|
33
|
35%
|
21
|
30
|
41%
|
|
COMP
|
4
|
4
|
50%
|
3
|
5
|
63%
|
3
|
5
|
63%
|
|
COOP
|
25
|
38
|
60%
|
22
|
41
|
65%
|
38
|
48
|
76%
|
|
NINTa
|
3868
|
2497
|
61%
|
3868
|
2497
|
61%
|
5259
|
1106
|
83%
|
|
COMPa
|
69
|
54
|
44%
|
36
|
87
|
71%
|
68
|
55
|
44%
|
|
COOPa
|
248
|
443
|
64%
|
186
|
500
|
73%
|
154
|
537
|
77%
|
- Note: COOP and COMP predictions from NAUTICA were collapsed into the general ``interaction” (INT) category for the comparison. Upper: no calibration. Lower: with calibration (also marked with a). Calibration is done with the same procedure as the general NAUTICA recall analysis (Table 2)