Derive and transform features from cleaned datasets for CRISP-DM workflows.
npx skills add https://github.com/thbraet/claude-template --skill construct-data
Please help me install this Agent Skill. Skill: construct-data Source: https://github.com/thbraet/claude-template/tree/main/skills/construct-data Command: npx skills add https://github.com/thbraet/claude-template --skill construct-data
Machine learning pipelines often struggle with consistent feature engineering and proper documentation. This skill standardizes the process of deriving, transforming, and recording features to support reproducible CRISP-DM workflows.
Load your cleaned dataset, run the feature construction pipeline, and generate both a Jupyter notebook and a CRISP-DM summary document.
High-intent search queries and answers about installing and using this skill.
Document feature engineering by generating a working Jupyter notebook at notebooks/3.3-construct-data.ipynb and a structured CRISP-DM summary report at docs/crisp-dm/3-data-preparation/3.3-construct-data.md, ensuring all transformations are versioned and auditable.
Standardize derived features and transformations by applying a reproducible feature engineering pipeline within the CRISP-DM 3.3 data preparation phase, which plans, derives, and records synthetic records and attributes into a summary document.
Generate a feature engineering report alongside a notebook by running the construction pipeline on a cleaned dataset, which automatically outputs both a reproducible notebook and a structured markdown summary document.
Yes, the feature construction process supports synthetic record generation. It allows you to derive new attributes, create derived measurements, and generate synthetic records when appropriate for your modeling needs.
You need a cleaned dataset as the input format before starting feature construction. The pipeline processes this cleaned data to plan, derive, and transform features for your machine learning models.
Yes, all feature construction steps are versioned and auditable. The pipeline enforces reproducibility by recording every transformation and derivation in the generated notebook and summary report.
qoobaGenerate tweaktune pipelines for synthesizing LLM training datasets via interactive Q&A flows.
cardmagicClassify text sentiment and spam using pre-trained CLI models.
thbraetApply type casting, column ordering, and deterministic splits to finalize datasets for modeling.
thbraetAutomate data cleaning for CRISP-DM 3.2 workflows with notebooks and summaries.
