Automated hyperparameter selection for the PC algorithm

The PC algorithm infers causal relations using conditional independence tests that require a pre-specified Type I α level. is however unsupervised, so we cannot tune traditional cross-validation. We therefore propose AutoPC, fast procedure optimizes directly for user chosen metric. in particular force to double check its output by executing second run on the recovered graph. choose final as one which maximizes stability between two runs. AutoPC consistently outperforms state of art across multiple metrics.