pershout

This is somewhat of a catchall repository for various experiments over the years on outlier detection via persistent homology. This includes:

using a similarity metric between time series to evaluate a distance matrix for a data set and then constructing the 1-D topological skeleton using a minimal spanning tree from which outliers are the least connected data points
fitting Gaussian processes to time series and then evaluating the distance matrix for a data set using the Wasserstein metric and constructing the persistence diagram from this
constructing persistence diagrams for individual time series and then evaluating a distance matrix for a data set and determining outliers from this

Name		Name	Last commit message	Last commit date
Latest commit History 12 Commits
Deconstructing the WTFs.ipynb		Deconstructing the WTFs.ipynb
Gaussian Process Persistent Homology.ipynb		Gaussian Process Persistent Homology.ipynb
LICENSE		LICENSE
LSST ISSC Hackathon Notes.md		LSST ISSC Hackathon Notes.md
README.md		README.md
TDA for time series.ipynb		TDA for time series.ipynb
TDA_for_time_series_limiting_cases.ipynb		TDA_for_time_series_limiting_cases.ipynb
data.csv		data.csv
pershout.py		pershout.py

Provide feedback