As metioned before Yummy uses centralized features selection logic in th backend implementations, thus adding new data source is very simple. Moreover you can combine multiple data sources to select required features.

Below you have list of sources available for each backend.

Sources/Backends Spark Polars Dask Ray
Iceberg Source ✔️
Delta Source ✔️ ✔️ ✔️ ✔️
Parquet Source ✔️ ✔️ ✔️ ✔️
CSV Source ✔️ ✔️ ✔️ ✔️

Table of contents