Search Algorithms and Loss Functions for Bayesian Clustering
Compute the Bell Number
Canonicalize Cluster Labels
CHIPS Partition Greedy Search
Latent Structure Optimization Based on Draws
Enumerate Partitions of a Set
Enumerate Permutations of Items
Compute Partition Loss or the Expectation of Partition Loss
Heatmap, Multidimensional Scaling, Pairs, and Dendrogram Plotting for ...
Compute an Adjacency or Pairwise Similarity Matrix
salso: Search Algorithms and Loss Functions for Bayesian Clustering
SALSO Greedy Search
Summary of Partitions Estimated Using Posterior Expected Loss
Threshold CHIPS Output
The SALSO algorithm is an efficient randomized greedy search method to find a point estimate for a random partition based on a loss function and posterior Monte Carlo samples. The algorithm is implemented for many loss functions, including the Binder loss and a generalization of the variation of information loss, both of which allow for unequal weights on the two types of clustering mistakes. Efficient implementations are also provided for Monte Carlo estimation of the posterior expected loss of a given clustering estimate. See Dahl, Johnson, Müller (2022) <doi:10.1080/10618600.2022.2069779>.