The goal of dendroMix is to summarize and select mixture models via dendrograms constructed from overfitted latent mixing measures. The methods bridge hierarchical clustering and mixture modeling, providing consistent selection of mixture components and interpretable hierarchical structure discovery. Includes functions for simulation, visualization, and applications such as single-cell RNA-seq analysis.
If you find the dendroMix R package or any of the source code in this repository useful for your work, please cite:
Dat Do, Linh Do, Scott A. McKinley, Jonathan Terhorst, and XuanLong Nguyen. Dendrogram of mixing measures: Hierarchical clustering and model selection for finite mixture models. ArXiv, 2403.01684 (2024).
# install.packages("remotes")
remotes::install_github("ldo3/dendroMix")