analysis-conventions

Standardizing data analysis project structures for reproducibility requires enforcing a unified framework for script execution and environment management, specifically utilizing uv for dependency control to ensure deterministic pipeline outputs across engineering workflows.

Managing exploratory notebooks versus deterministic pipeline scripts requires defining clear workflow boundaries, separating interactive Marimo notebooks for exploration from dedicated deterministic scripts to maintain reproducibility and clarity in data science projects.

Yes, you need uv to manage dependencies in your data science workflow because this convention enforces uv-based execution for environment standardization, ensuring all script dependencies are explicitly controlled and reproducible across different execution contexts.

Handling temporary scratchpad files without cluttering your analysis project requires establishing an artifact hygiene protocol that manages temporary files systematically, preventing workspace clutter while maintaining auditability of calculated values and data sources.

Documenting calculated metrics and data sources in data analysis pipelines is required to resolve project inconsistencies, providing explicit source references for formulas to maintain auditability and ensure reproducibility within engineering workflows.

Marimo notebooks are designated for exploratory analysis rather than deterministic pipeline execution, requiring clear workflow patterning that separates interactive Marimo notebooks from deterministic pipeline scripts to maintain reproducibility and structural clarity.