TLDR Tech / Data
Machine Learning with ADBC, DuckDB & XGBoost: No Pandas, Just Pure Arrow Tables
This walkthrough shows a practical ML workflow where ADBC pulls data from DuckLake as Arrow tables directly into XGBoost, with SQL handling preprocessing and PyArrow handling the train/test split instead of Pandas or scikit-learn. The main point is that ADBC and Arrow can simplify and standardize database-to-ML pipelines with fewer libraries and a cleaner, more portable workflow.
Read the full story