This is a python implementation of:
Studies in Astronomical Time Series Analysis. VI. Bayesian Block Representations: Jeffrey D. Scargle, Jay P. Norris, Brad Jackson, James Chiang - Link
It is based upon the example MATLAB code and the underlying mathematical description of the algorithm in the paper. The code has been tested in Python 2.7 and 3.5. It is still in development as while the analysis is easy to perform on a single time-series, it is not 'user-friendly' to deploy it on multiple timeseries at this time.
This is test checks everything is running as expected but doesn't expose the specifics of how to run a Bayesian blocks analysis
import BayesBlocks as bb%matplotlib inline
subDF, myBlocks = bb.testBayes()Input data processed ready for Bayesian Block calculation ...
Using a FAP of 0.050000 equates to a changepoint prior of 7.443134
Block fitness function assessed ...
Changepoints recovered ...
Post processing complete...
==============
Analysis of input data has found that it is optimally segemented into 14 blocks
[ 0 1751 3001 3247 4577 5013 6565 8172 9393 10813 11635 12114
16867 20474]
