This repository contains implementation for paper : A Condiional Independence Test in the Presence of Discretization [arXiv]
DCT is a conditional independence test specifically designed for the scenario that only discretized version of variables available. Specifically, DCT tries to recover the covariance matrix
run the code
conda env create -f environment.yml
Then you will have a conda environment named 'causal'. You can further activate the environment by running
conda activate causal
We provide two examples running the test in example_to_use,ipynb
and running the PC algorithm with DCT as the test in example_to_use_pc.ipynb
. Our core algorithm is implemented at causal_learn.causallearn.utils.DisTestUtil.py
.