About these benchmark studies

These six applications use public benchmarks. Each chart compares default Causilo with the strongest recorded configuration of each conventional model family. The recorded results retain all 51 datasets and original scores, including weaker Causilo results.

Task results use the September 14 snapshot. ROC AUC differences are percentage points. RMSE and log-loss differences are relative error changes. These benchmark means do not establish statistical significance or business outcomes.

Try your own prediction task

Start with known examples and a target column. Evaluate predictions on held-out rows before using them in your workflow.

Start Building