Shape-based Clustering of Multivariate Longitudinal Data
We extend shape-based clustering for longitudinal data to the novel application of microbiome taxa composition with zero inflation. We find interesting longitudinal profiles for some clusters of subjects.
Most of the analysis is done in Python; some summary and EDA is done in R. Below is a description of some files:
mvtraj-group-paper.pdf : our class paper
mvtrajectories.py : our methods
mb-trajectories.ipynb : k-means
simulations.ipynb : a simulation study
bootstrap_final_clusters.R : some bootstrapping to study significance
This is a course project. Python functions may error out for unexpected input types.