1. Mixed-data clustering and visualization- The aim of the project is the development of utilities for clustering subjects and variables of mixed data types. The main purpose is the generation of a mixed-data heatmap.
2. bclust- Bayesian clustering with variable selection is a fully automatic clustering method, designed for low-sample-size-high-dimensional situations. This might be regarded as a high-dimensional variant of the mclust package.
Development Status : 5 - Production/Stable [Filter]
4. ziLRhClust: hclust for zeroinflated data- Algorithm to process hierarchical clustering, based on log-likelihood ratio, considering standard Gaussian or zero-Inflated Gaussian data. The agglomeration method is also the log-likelihood-based dissimilarity between clusters.
6. Clustering using convex fusion penalties- An R/C++ implementation of the clusterpath algorithm described in Hocking et al. 2011, for robust convex clustering using sparsity-inducing fusion penalties.
7. Local Depth- The package "localdepth" contains functions for the evaluation of some Depth functions and their corresponding local versions, namely Local Depths. Simplicial, Ellipsoid and Mahalanobis Depths are implemented for general dimension.
8. Bayesian Biclustering- This package uses a Bayesian spike-and-slab model to construct bidendrograms using log posterior as the natural distance defined by the model and calculates importance using log Bayes factor.