As evidenced by the results shown in Table 3, using each of the hybrid algorithms produced results better than those obtained by arFCM, at least in 62% of applications. In the case of minimizing the objective function, the WFCM algorithm provided the best results.
Table 3
Percentage of points over the range in which each proposed hybrid algorithm showed better performance than arFCM according to values of the objective function and cluster validity indices for
| Measure | |||||
|---|---|---|---|---|---|
| Algorithm | Objective function (J) | VPC | VPE | VXB,m | VK |
| SLFCM | 62 | 87 | 88 | 93 | 92 |
| CLFCM | 72 | 79 | 80 | 97 | 91 |
| ALFCM | 68 | 82 | 86 | 96 | 90 |
| WFCM | 84 | 65 | 67 | 81 | 76 |
| SOFMFCM | 68 | 62 | 62 | 77 | 76 |
| Measure | |||||
|---|---|---|---|---|---|
| Algorithm | Objective function (J) | VPC | VPE | VXB,m | VK |
| SLFCM | 62 | 87 | 88 | 93 | 92 |
| CLFCM | 72 | 79 | 80 | 97 | 91 |
| ALFCM | 68 | 82 | 86 | 96 | 90 |
| WFCM | 84 | 65 | 67 | 81 | 76 |
| SOFMFCM | 68 | 62 | 62 | 77 | 76 |