Add clustering methods into the benchmark games. For two or more clustering algos, conduct basically a feature selection game with remove and refitting of the clusters. The value of the game can be a sensible loss function with the BIC information criterion being a sensible default.
[x] Implement base unsupervised clustering game class
Add clustering methods into the benchmark games. For two or more clustering algos, conduct basically a feature selection game with remove and refitting of the clusters. The value of the game can be a sensible loss function with the BIC information criterion being a sensible default.