Table 14:
Adjusted Rand index, F1 score, along with estimated number of clusters and features of adaptive iGecco+ and iClusterPlus for high-dimensional mixed multi-view data of Table 4 and 5 in Section 4.2 when the number of clusters and features are not fixed but estimated based on the data. We only include estimated number of clusters for iClusterPlus as there is no tuning parameter for the number of selected features.
| S3 | S4 | S5 | S6 | |||||
|---|---|---|---|---|---|---|---|---|
| ARI | # of Clusters | ARI | # of Clusters | ARI | # of Clusters | ARI | # of Clusters | |
| Algorithm 17 + 15 (BIC) | 0.97 (7.8e-3) | 3.00 (0.0e-0) | 0.97 (1.3e-2) | 3.10 (1.0e-1) | 1.00 (0.0e-0) | 3.00 (0.0e-0) | 1.00 (0.0e-0) | 3.00 (0.0e-0) |
| Algorithm 17 + 16 (SS+BIC) | 0.93 (4.1e-2) | 2.90 (1.0e-1) | 0.89 (5.5e-2) | 2.80 (1.3e-1) | 0.99 (1.2e-2) | 3.10 (1.0e-1) | 0.82 (8.1e-2) | 4.90 (9.1e-1) |
| iCluster+ | 0.53 (8.1e-2) | 2.70 (3.0e-1) | 0.72 (5.3e-2) | 3.50 (3.1e-1) | 0.63 (2.4e-2) | 3.50 (2.2e-1) | 0.60 (1.4e-2) | 3.30 (1.5e-1) |
|
| ||||||||
| S3 | S4 | S5 | S6 | |||||
| F1-score | # of Features | F1-score | # of Features | F1-score | # of Features | F1-score | # of Features | |
|
| ||||||||
| Algorithm 17 + 15 (BIC) | 0.93 (1.1e-2) | 30.00 (1.1e-0) | 0.95 (1.9e-2) | 31.60 (7.2e-1) | 0.99 (6.3e-3) | 30.50 (4.0e-1) | 0.99 (6.5e-3) | 30.90 (4.1e-1) |
| Algorithm 17 + 16 (SS+BIC) | 0.93 (1.2e-2) | 29.60 (1.2e-0) | 0.96 (1.5e-2) | 29.40 (6.5e-1) | 0.99 (6.3e-3) | 30.50 (4.0e-1) | 0.99 (6.0e-3) | 30.10 (3.8e-1) |