Calculation of per-cluster entropy

Description

Provides the calculation of per-cluster entropy, equivalent to

Entropy(k) = ∑_{i \in C_k} \log (τ_{ik})

where τ_{ik} is the conditional probability of gene i belonging to cluster k and C_k corresponds to the set of indices of genes attributed to cluster k.

Usage

1
clusterEntropy(probaPost)

Arguments

probaPost

Matrix containing the conditional probabilities of belonging to each cluster for all observations

Value

Entropy per cluster

Author(s)

Cathy Maugis-Rabusseau

Examples

1
2
3
4
## Generate artificial matrix of conditional probabilities for K=5 clusters
tmp <- matrix(runif(100*5), nrow=100, ncol=5)
probaPost <- tmp / rowSums(tmp)
clusterEntropy(probaPost)