PyMC now includes experimental support for Linear, Gaussian state space time series models through the pymc_experimental.statespace module. This is a game-changer for anyone working with time series data, offering a more streamlined and efficient way to handle Linear Gaussian state space models.
State-space models are incredibly versatile, allowing you to represent various time series models, including SARIMAX, VARMAX, and more. However, they have historically been cumbersome to work with in PyMC, requiring extensive setup and manual handling of hidden states.
With this new addition, you can easily work with linear Gaussian state space models directly in PyMC for tasks like parameter estimation, hidden state inference, missing data interpolation, forecasting, and more. This will save you time and make your time series analysis more accessible and powerful.
If you’re interested in diving deeper, checkout this post:
There are examples of usage and a tutorial on creating custom state space models linked in the post. You can get started by installing it with pip install pymc-experimental in your favorite (conda-installed) PyMC environment.
A big shoutout to @jessegrabowski for making this possible.
State space models are an essential tool in various fields, and this update makes them more accessible and efficient for data scientists and researchers. Give it a try, and let us know your thoughts and feedback.
Happy modeling!
Don’t forget to give this post a thumbs-up and share it with your network!