Table 1: Confusion matrices, size groups, estimates of the parameters, variability measures (between-groups and within-groups) and F -statistic for the EM algorithm (soft-movMF, hard-movMF, stochastic EM) and dynamic clusters algorithm (DC), for the sample size of 20 and concentrations equal to 5 or 10 (*: the results for the other three methods are equal).
| k |
θ (°) |
Algorithm |
Group |
Conf. matrix |
ni
|
 |
 |
Bet./With. |
F |
| 1 |
2 |
| 5 |
30 |
Soft-movMF |
1 |
7 |
3 |
14 |
15.5 |
46.4 |
23.8/18.5 |
23.3 |
| 2 |
7 |
3 |
6 |
18.1 |
| Hard-movMF |
1 |
7 |
3 |
14 |
16.6 |
48.1 |
26.2/20.0 |
17.1 |
| 2 |
7 |
3 |
6 |
20.3 |
| DC |
1 |
6 |
4 |
13 |
17.3 |
46.4 |
22.2/23.3 |
17.1 |
| 2 |
7 |
3 |
7 |
18.8 |
| 90 |
Soft-movMF |
1 |
6 |
4 |
6 |
14.4 |
77.5 |
32.0/18.8 |
30.6 |
| 2 |
0 |
10 |
14 |
5.4 |
| Hard-movMF |
1 |
8 |
2 |
8 |
9.2 |
77.9 |
31.1/20.0 |
28.0 |
| 2 |
0 |
10 |
12 |
7.3 |
| Stochastic EM
|
1 |
6 |
4 |
6 |
16.2 |
78.9 |
34.3/20.0 |
30.9 |
| 2 |
0 |
10 |
14 |
5.6 |
| DC |
1 |
9 |
1 |
12 |
6.1 |
75.4 |
27.6/20.0 |
24.9 |
| 2 |
3 |
7 |
8 |
10.0 |
| 150 |
Soft-movMF* |
1 |
10 |
0 |
10 |
5.9 |
127.3 |
70.3/20.0 |
63.3 |
| 2 |
0 |
10 |
10 |
12.5 |
| 10 |
30 |
Soft-movMF |
1 |
4 |
6 |
4 |
77.5 |
42.3 |
34.3/19.4 |
31.8 |
| 2 |
0 |
10 |
16 |
14.2 |
| Hard-movMF |
1 |
8 |
2 |
13 |
12.8 |
33.9 |
15.6/20.0 |
14.1 |
| 2 |
5 |
5 |
7 |
32.6 |
| DC |
1 |
7 |
3 |
11 |
14.4 |
33.4 |
13.5/20.0 |
12.2 |
| 2 |
4 |
6 |
9 |
20.9 |
| 90 |
Soft-movMF* |
1 |
10 |
0 |
10 |
10.9 |
90.0 |
72.9/20.0 |
65.6 |
| 2 |
0 |
10 |
10 |
18.2 |
| 150 |
Soft-movMF* |
1 |
10 |
0 |
10 |
10.9 |
142.4 |
163.3/20.0 |
147.0 |
| 2 |
0 |
10 |
10 |
20.4 |