Baseline

Here, we describe the baseline models and evaluation metrics used.

Models

Examples to run baselines can be found in notebooks/eval_pipeline.ipynb. Follow installation and (more) runtime instructions of each baseline in the provided github links.

• PCMCI+: jakobrunge/tigramite

• FPCMCI: lcastri/fpcmci

• VARLiNGAM: cdt15/lingam

• DYNOTEARS: mckinsey/causalnex

• Neural GC: iancovert/Neural-GC

• CUTS+: jarrycyx/unn

• TSCI: KurtButler/tangentspace

Metrics

We use AUROC and AUPRC scores to evaluate the accuracy of discoverd summary graph.