Construct machine learning pipelines using skrub DataOps graphs.
npx skills add https://github.com/vathymut/copilot-skills --skill build-ml-pipeline
Please help me install this Agent Skill. Skill: build-ml-pipeline Source: https://github.com/vathymut/copilot-skills/tree/main/.github/skills/build-ml-pipeline Command: npx skills add https://github.com/vathymut/copilot-skills --skill build-ml-pipeline
This Skill provides a structured approach to building machine learning pipelines, ensuring they are declarative and maintainable.
build-ml-pipeline "data_dir=/path/to/data" "predict_grid=/path/to/predict_grid" "history_source=/path/to/history_source" " learner=HistGradientBoostingRegressor"
High-intent search queries and answers about installing and using this skill.
You build a machine learning pipeline by defining a skrub DataOps graph that structures data processing and feature engineering. This approach ensures your pipeline remains declarative, maintainable, and clearly structured for predictive modeling tasks.
Yes, you can use sklearn estimators with skrub for feature engineering. The Skill differentiates between stateless functions and stateful estimators, allowing you to integrate sklearn components seamlessly within the skrub DataOps graph for predictive modeling.
The best way to handle multi-table inputs is by using skrub joiners within your machine learning pipeline. This allows you to declaratively merge multiple data sources while maintaining a structured DataOps graph for downstream feature engineering.
You tune hyperparameters by incorporating meta-estimators and utilizing hyperparameter knobs within the skrub DataOps graph. The skore library is then used alongside sklearn to evaluate the pipeline construction and ensure optimal model performance.
Yes, you need to ensure early X marker placement for proper feature engineering. This Skill explicitly handles the correct placement of the X marker within the skrub DataOps graph to separate training inputs from targets effectively.
vathymutSet up a Python data science stack with scikit-learn, PyTorch, and Polars.
0xZKnwTrain sklearn models with walk-forward validation on OHLCV data.
SnakeORefactor scikit-learn ML code into Pipelines with ColumnTransformer and deterministic random_state.
TeamSparkAIBuild no-code MCP servers using YAML graphs with JSONata and JSON Logic.
