Table 2.
AICc and BIC scores of the best partitioning scheme found by different algorithms on each dataset
| |
AICc |
BIC |
||||
|---|---|---|---|---|---|---|
| Dataset |
Greedy |
Relaxed clustering |
Strict clustering |
Greedy |
Relaxed clustering |
Strict clustering |
| (AICc) | (ΔAICc) | (ΔAICc) | (BIC) | (ΔBIC) | (ΔBIC) | |
| Ward_2010 |
103258 |
-34 |
-61 |
104877 |
-294 |
-606 |
| Wainwright_2012 |
473537 |
-7 |
-59 |
477322 |
-73 |
-663 |
| Pyron_2011 |
154838 |
-42 |
-173 |
156039 |
-177 |
-383 |
| Li_2008 |
252583 |
-6 |
-242 |
254327 |
-183 |
-769 |
| Leavitt_2013 |
424129 |
-216 |
-757 |
426143 |
-837 |
-3176 |
| Kaffenberger_2011 |
120020 |
-6 |
-75 |
121452 |
-62 |
-150 |
| Irisarri_2012 |
214655 |
-41 |
-187 |
216209 |
-152 |
-1151 |
| Hackett_2008 |
1830824 |
-356 |
-1442 |
1837230 |
-964 |
-6362 |
| Fong_2012 |
276517 |
-254 |
-1508 |
278400 |
-900 |
-2129 |
| Endicott_2008 | 66966 | -90 | -479 | 70139 | -455 | -752 |
The greedy algorithm performed best in all cases, as expected, and the AICc/BIC score is shown for each run with that algorithm. The relaxed clustering algorithm typically performed almost as well as the greedy algorithm, and always performed better than the strict clustering algorithm. ΔAICc or ΔBIC scores are shown for the clustering algorithms, and represent the difference in AICc or BIC score from the greedy algorithm.