User Guide – Automated Analysis

FoodSpec supports highly automated “one-click” analysis in two paths:

Group B: CLI Auto-Analysis + Publish

• Why: Scriptable, reproducible runs for batch use.
• How:

  # Run protocol end-to-end
  foodspec-run-protocol \
    --input examples/data/oils.csv \
    --protocol examples/protocols/oil_basic.yaml \
    --output-dir runs/auto_oil_basic \
    --auto --report-level standard
  
  # Auto-generate narrative and figure panels
  foodspec-publish runs/auto_oil_basic/<timestamp> --fig-limit 6
  

• For multi-input/harmonized runs: add multiple --input flags (e.g., oils + chips).
• For HSI: use hsi_segment_roi protocol and the HSI example data.
• Outputs: run bundle with report.txt/html, figures/, tables/, metadata.json, index.json, run.log, and optionally models/ (frozen pipelines).
• Dry-run: use --dry-run to validate/estimate without executing (helpful before large HSI/multi-input runs).

Tips for best automation

• Always run validation via the protocol’s validation block; fix blocking errors before proceeding.
• Let protocols choose validation strategy by default; they auto-reduce CV folds when classes are tiny.
• Use HDF5 for multi-instrument/HSI to retain harmonization metadata.
• For repeated runs, organize CLI outputs under a project folder.

Cross-links: cli_guide.md, cookbook_rq_questions.md, validation_strategies.md.