PowerForecast Docs
Back to Wizard
This page belongs to the full documentation. You don't need it to plan: back to the essentials.

Demos

Every demo uses real public data (provenance: forecast-datasets/MANIFEST.md). Each lives in its own folder of forecast-demos/.

The reference demo: US retail sales

04 · US retail sales: 2026 budget, reality check, 2027 re-forecast

A retail analytics team forecasts monthly sales of 20 kinds of business (US Census Monthly Retail Trade, millions of USD) for the 2026 budget, from 11 years of history. In September 2026, with eight months of real 2026 sales published, it checks how good the budget forecast was, then re-forecasts to December 2027.

FileWhat
input/spec.budget.xlsxthe workbook: Setup + Data (20 series × 132 months, 2015 – 2025)
input/spec.reforecast.xlsxthe same with 2026 actuals to August (140 months)
SCENARIO.mdthe business story, the data, what to show
EXPECTED_RESULTS.mdthe engine's results, and the budget forecast against the real 2026 months
DEMO_SCRIPT.mdstep by step, what to click and what to say
_generate.py · _run.py · _check.pyrebuild the specifications, run them through the engine, check against the actuals

AI-assistance demos

Short demos of the per-step AI assist panel: paste plain language, get the fields drafted.

#DemoData
01Retail demand, monthlyAirPassengers, 144 months
02Insurance claims, weeklyUS weekly jobless claims, 261 weeks
03Contact-centre workload, dailydaily births, 365 days

How to run them: forecast-demos/README.md.

Rebuilding a demo workbook

python -I forecast-demos/04-us-retail-sales-budget/_generate.py
node forecast-wizard/tools/make-demo-kit.mjs forecast-demos/04-us-retail-sales-budget/spec.budget.json
node forecast-wizard/tests/demo-kits.test.mjs