This repository provides an implementation of the bottom-up procedure defined in Optimal multiple change-point detection for high-dimensional data and compares its performances with the inspect method from High-dimensional changepoint estimation via sparse projection and where an implementation can be found in inspect method.
The procedure is implemented in mutichange.py and the notebook performances.ipynb explains how we get to the result of the paper. The csv files are the data obtained to draw the illustrations and they can be redrawn with the notebook savingPNG.ipynb without redoing all the computations.