Important changes
- string feature extractor now uses a proper categorical encoder for lightgbm and xgboost: it should speed up the training a lot
- bootstrap job got a lot of performance improvements for large datasets: we've seen reductions from 30m to 2m in practice.
What's Changed
- migrate to flink 1.15 and scala 2.13 by @shuttie in #432
- Update jedis to 4.2.3 by @scala-steward in #430
- support index category encoding for string extractor by @shuttie in #435
- bump http4s to m32 by @shuttie in #436
- Update scalafmt-core to 3.5.3 by @scala-steward in #437
- fix broken logo, add doc link by @shuttie in #440
- make index encoding default for categorical features by @shuttie in #441
- Update scaffeine to 5.2.0 by @scala-steward in #442
- Bootstrap performance work by @shuttie in #444
- Update flink-scala-api to 1.15-2 by @scala-steward in #443
Full Changelog: 0.3.0...0.3.1