Comparison of causal-learn to Tetrad for analyzing the NASA Airfoil data.

I did a comparison of causal-learn to Tetrad on the task of analyzing the NASA Airfoil Self-Noise data (from the UCI repository) from a causal perspective. This is experimental data, so we know something of the ground truth, though not the full model. It presents various challenges to causal modeling; suggestions are made as to features to add to causal-learn (that are available currently in Tetrad) to help analyze it.

The document may evolve, so I give an Overleaf link that if successful will allow you to view the PDF. The PDF was created by Claude after a long discussion of how to analyze the data, with some features added to Tetrad specifically to analyze this and similar data.

https://www.overleaf.com/read/zvgdxhxgrptg#436652

This is in service of the goal of making it so public software can analyze real data successfully. I added the (causal-learn) FCI Fisher Z alpha 0.05 model recommended by causal-learn.com.

Start by reading the linked Overleaf document and reviewing the causal-learn FCI Fisher Z alpha 0.05 model mentioned in the issue. The comparison discusses possible features from Tetrad, but no specific file, test, or requested change is identified, so the completion criteria need clarification before implementation.