Bayesian Hierarchical Invariant Causal Predictors
src folder contains the code for the BHIP algorithm. The code is written in Python.
bus_dwelling_time.ipynb is a notebook that contains the bus dwelling problem as well solutions using BHIP with non-centered parametrization, horseshoe or spike and slab priors as well as ICP.
experiments.py and computational_study.py have the scripting code to run the experiments relative to the quantitative study and the computational study, respectively.
Finally example_environment_split_school.ipynb is a notebook that contains the example of the school problem as well as solutions using BHIP with non-centered parametrization. Horseshoe, spike and slab priors as well as ICP solutions are in school_spike_and_slab.ipynb.