Opinionated machine learning pipelines for training and inference.
To do
To do.
Declare any dependencies in requirements.txt for pip installation.
To install them, run:
pip install -r requirements.txt
You can run your Kedro project with:
kedro run
Candidate options: xgboost, lightgbm and catboost. We compared the three packages based on 3 criteria:
- number of releases in 2022-2023
- number of dependencies (important when working with containers)
- package size
Based on those criteria, we had the following result:
| xgboost | lightgbm | catboost | |
|---|---|---|---|
| # releases | 14 | 7 | 9 |
| dependencies | 2 (numpy/scipy) | 2 (numpy/scipy) | 18 |
| size (MB) | 6.2 | 5.1 | 90 |