Table 5.
Summary of results for Simulation II: The table reports, for each category level of each design feature, the average of the ARI values for the all tested combinations of method (LCM, K-means, K-median) and criterion for choosing K (AIC2, AIC3, BIC, CH, MRPC).
| Design feature | Level | LCM (AIC2) | LCM (AIC3) | LCM (BIC) | K-means (CH) | K-means (LCM_AIC3) | K-median (MRPC) | K-median (LCM_AIC3) |
|---|---|---|---|---|---|---|---|---|
| Sample size | N = 100 | .8043 | .8069 | .7744 | .7159 | .7809 | .8038 | .8166 |
| N = 200 | .8104 | .8287 | .8178 | .7165 | .7966 | .8175 | .8315 | |
| N = 400 | .8175 | .8424 | .8424 | .7239 | .8081 | .8229 | .8382 | |
| Number of clusters | K = 2 | .8942 | .9273 | .9285 | .8450 | .9026 | .8819 | .9100 |
| K = 3 | .8522 | .8693 | .8593 | .7883 | .8624 | .8633 | .8684 | |
| K = 4 | .8186 | .8282 | .8128 | .6836 | .8015 | .8231 | .8334 | |
| K = 5 | .7610 | .7693 | .7445 | .6571 | .7295 | .7669 | .7822 | |
| K = 6 | .7276 | .7359 | .7126 | .6199 | .6801 | .7386 | .7499 | |
| Number of variables | V = 6 | .7095 | .7055 | .6810 | .6128 | .6469 | .6799 | .6952 |
| V = 9 | .8301 | .8519 | .8341 | .7178 | .8346 | .8523 | .8681 | |
| V = 12 | .8926 | .9206 | .9195 | .8257 | .9041 | .9121 | .9229 | |
| Cluster sizes | Equal | .8197 | .8240 | .8035 | .7257 | .8238 | .8342 | .8339 |
| 60% | .8043 | .8311 | .8262 | .7436 | .7675 | .8087 | .8333 | |
| 10% | .8081 | .8229 | .8049 | .6870 | .7943 | .8013 | .8190 | |
| Error level | 5% | .9211 | .9270 | .9243 | .9084 | .9191 | .9287 | .9317 |
| 10% | .8247 | .8419 | .8275 | .7107 | .8056 | .8307 | .8432 | |
| 15% | .6864 | .7091 | .6828 | .5373 | .6610 | .6848 | .7114 | |
|
| ||||||||
| Overall | .8107 | .8260 | .8115 | .7188 | .7952 | .8147 | .8288 | |