By Andrew Noble
This repo contains the Python files used to demonstrate the utility of a Principal Component Analysis (PCA) as a simple and scalable first step in searching for anomalous behavior in large spatiotemporal data sets. Two such data sets, one on the San Francisco bike share program and the other on UK measles outbreaks following World War II, are analyzed here. Code in the bikeshare directory reproduces the results discussed on this webpage. Code in the measles directory reproduces the results discussed on this webpage. Code in the util directory is used by both bikeshare and measles.
- Python (numpy, scipy, pandas, pickle, cartopy, matplotlib, pylab)
Clone the repo.
git clone https://github.com/andrewenoble/net-detect.git
cd net-detect
From the net-detect directory, decend into either the measles or bikeshare directory. Further usage instructions can be found there in another README.md file.
This work is support by an NSF INSPIRE award from the National Science Foundation.