Supercharging Deduplication in pandas Using RAPIDS cuDF

Originally published at: Supercharging Deduplication in pandas Using RAPIDS cuDF | NVIDIA Technical Blog

A common operation in data analytics is to drop duplicate rows. Deduplication is critical in Extract, Transform, Load (ETL) workflows, where you might want to study the latest records, find the first time a key appears, or remove duplicate keys completely from your data. Which rows you keep and the row ordering both impact downstream…