EXP-008 / question
Technical question
How can forecasting move from passive prediction to active scenario planning?
EXP-008 / method
Method and workflow
- Load SQL or CSV data into a structured forecasting workflow.
- Run Prophet trend, holiday, and seasonality modeling with a moving-average baseline.
- Apply scenario adjustments using elasticity and media-response logic.
- Visualize baseline vs scenario outcomes in Streamlit and Plotly.
- Export results as CSV or Markdown reports.
SQL / CSV
ETL
Prophet model
scenario engine
explainability
dashboard
exports
EXP-008 / evidence
Evidence of work
Business framing
The README frames the core goal as testing assumptions, quantifying tradeoffs, and preparing for volatility.
Scenario engine
The architecture includes elasticity adjustments for pricing, ad spend, and promotions.
Validation
The technical design includes backtesting with MAE, RMSE, and SMAPE.
EXP-008 / stack
Technical stack
PythonStreamlitProphetSQLAlchemypandasPlotlySQLite
Open repository ↗
EXP-008 / limitations
Limitations and honesty check
- Real business use would need stronger data validation, domain-specific priors, and clear uncertainty intervals.
- Scenario assumptions should be calibrated against historical experiments or expert input.
- A production setup would require scheduled retraining, monitoring, and model versioning.
EXP-008 / next
Next improvements
- Add uncertainty bands and scenario confidence messaging.
- Add model comparison beyond Prophet and moving average.
- Add automated data quality checks before forecasting.
- Create a public demo with sample datasets.