Proposal demo: the monthly forecast-vs-actuals ritual of a residential pest control business, as an app. Demo figures throughout.
The flow it sells: close a month, see plan vs actual variance on the full P&L (sales, gross margin, net margin), re-forecast the open months with your own driver assumptions, sanity-check them against a SARIMAX statistical forecast, and download one formatted Excel report for the management meeting.
- One input: the closed month (sidebar), exactly like the closed-month switch in an Excel model.
- P&L plus drivers: recurring contracts x price, seasonal one-time work, cost ratios, marketing spend. Scenario sliders re-forecast the open months.
- Forecast check: SARIMAX(1,1,1)x(0,1,1,12) on 30 months of history with a confidence band, plus a 1-step-ahead backtest that answers "how much should you trust this model?"
- Excel out only: the app is the workplace, the report is the deliverable.
pip install -r requirements.txt
streamlit run app.py