BOCD

Python implementation of Bayesian Online Changepoint Detection for a Normal-Gamma model with unknown mean and variance parameters.

BOCD.py is a python script that estimates the run-length posterior distribution of the well-drilling nuclear magnetic response data using the Algorithm 1 mentioned in Bayesian Online Changepoint Detection.

Some useful resources for understanding and implementing the mentioned paper are:

• Murphy, K. P. (2007). Conjugate Bayesian analysis of the Gaussian distribution
• http://gregorygundersen.com/blog/2019/08/13/bocd/
• http://gregorygundersen.com/blog/2020/10/20/implementing-bocd/

New to Bayesian Statistics?

Check out some of these resources:

• Bayes Theorem explained by 3Blue1Brown
• Bayesian vs Frequentist Probability
• Beginner's guide to probability density
• Understanding prior and posterior predictive distributions