Multivariate T - Githubissues

multimeric commented 3 years ago

Introduces a pluggable prior/posterior config for multivariate Gaussian data, with sensible defaults. Note that this only works for scipy > 1.6.0, where they introduced the multivariate t PDF. The library will remind you to upgrade if you have an old version.
Adds a test for this new configuration, as well as for the univariate one
Adds a "dev" and "multivariate" setup extra, meaning that you can pip install bayesian_changepoint_detection[dev] for development work (currently this installs pytest), or pip install bayesian_changepoint_detection[multivariate] (enforces that you have a new enough scipy version for this new feature)

multimeric commented 3 years ago

Oops, a colleague just pointed out a huge flaw in my implementation. I'm putting this back to draft while I fix it.

multimeric commented 3 years ago

Well, I don't love the accuracy of it, but does seem to output a meaningful signal for multiple changepoints. I also copied a plot function used in the notebook into the module for others' use. Here is an example calculated using the multivariate test data:

hildensia / bayesian_changepoint_detection

Multivariate T #26