Introduction to gCastle
In this section, we’ll introduce gCastle (Zhang et al., 2021) – a causal discovery library that we’ll use in this chapter. We’ll introduce four main modules: models, synthetic data generators, visualization tools, and model evaluation tools. By the end of this section, you’ll be able to generate a synthetic dataset of chosen complexity, fit a causal discovery model, visualize the results, and evaluate the model using your synthetic data as a reference.
Hello, gCastle!
What is gCastle?
It’s an open source Python causal discovery library created by Huawei’s Noah’s Ark Lab. The library provides us with a comprehensive selection of modern causal discovery algorithms that include classics such as the PC algorithm, as well as cutting-edge gradient- or reinforcement-learning-based methods.
The repository (https://bit.ly/gCastleRepo) includes example notebooks, a list of currently available models,...
