sprcom 0.2

Bayesian spatial regression of communities

SpRCom

sprcom stands for Spatial Regression of Communities and is a statistical package designed to streamline the interpretation and modeling of very high dimensional binary and count-valued data. The underlying model assumes a low-dimensional latent structure via communities or clusters that leads to a parsimonious model. sprcom can also account for the dependence of these communities on covariates! A number of utility and plotting functions are included to help visualize your results. sprcom is a wrapper for a PyMC3 model and you can use any PyMC3 estimation method with it including Hamiltonian Monte Carlo and ADVI.

covariates, response, adjacency = load_data(...)
n_communities = 5
model = spatial_community_regression(covariates, response, adjacency,n_communities)
with model:
  trace = pm.sample()
...

We've included documentation to help you get up and running. Check out the florabank1-tutorial notebook for more details!

For questions or comments please contact Christopher Krapu at [e-mail].