This is an open-source implementation of the VLDB 2025 paper, Suna: scalable causal confounder discovery in relational datasets.
For a runnable example, open the Jupyter notebook at dataprofile/demo.ipynb.
It walks through the end-to-end Suna workflow on a small sample dataset.
This release assumes equi-joins and does not perform automatic data integration.
We’re currently working on an end-to-end dataset search system to support richer join conditions.
Please drop me a line at jl6235@columbia.edu if you would like to collaborate with the DAPLab