Shao17

In this repository, we reproduce (partially) the results of the following paper:

• W. Shao, J. Rougier, A. Paris, F. Devienne and M. Viste, "One-to-One Matching of RTT and Path Changes," 2017 29th International Teletraffic Congress (ITC 29), Genoa, 2017, pp. 196-204. https://arxiv.org/pdf/1709.04819.pdf.

We reuse the authors code and dataset (WenqinSHAO/rtt) written in R and Python, and add the HDP-HMM in the benchmark. We are able to reproduce exactly the same results as in the paper.

Modifications

• external/rtt: copy of WenqinSHAO/rtt modified to work with Python 3.

Usage

Tested with Julia 1.4+, Python 3.6+, R 4.0+.
Please see Overview/Usage first.

# R dependencies
R -e "install.packages('changepoint', repos = 'http://cran.us.r-project.org')"
R -e "install.packages('changepoint.np', repos = 'http://cran.us.r-project.org')"

# Python dependencies
pip install munkres scikit-learn

# In Shao17/
julia --project=notebooks/ -e 'import Pkg; Pkg.instantiate()'
jupyter lab

Notebooks

Name	Description
Benchmark	Benchmark of the HDP-HMM and several classical changepoint detection methods.