PowerForecast documentation

PowerForecast forecasts time series: sales, demand, workload, claims, transactions, revenue. You give it your history, for example one column per product, store or department in an Excel sheet, and say how far ahead you need to see. It studies each series (trend, seasons, irregularities), tries several forecasting methods on your own past, keeps the most accurate one and tells you, in plain words, whether you can rely on the result.

The wizard is in 5 languages (English, French, German, Spanish, Dutch) and has two themes (violet, amber).

Start here

Document	For	What's in it
PowerForecast in plain words	planners, managers	what a forecast is, the 4 steps, how to read the verdict, what to do when a point needs checking
Quick Start	new users	a forecast from an Excel workbook in 5 minutes
Concepts	everyone	series, time unit, horizon, typical error, likely range, methods, the checks behind the verdict

Use the wizard

Document	What's in it
Wizard Guide	every step: the Excel workbook, your data, how far ahead, review, running the forecast
User Guide	the Results page tab by tab, the files a run saves, reopening a run, changing a forecast, tips
Demos	the demos, each with its Excel workbook, scenario, expected results and demo script

Build, run and integrate

Document	What's in it
API Reference	`POST /jobs`, `GET /jobs/{id}`, `/result`, `/diagnostics`, `POST /forecast`, the JSON in and out, the envelope, errors
Developer Guide	code structure, the engine pipeline, the wizard and Results modules, tests, demo kits, adding a method
Installation (Hetzner)	server installation, containers, proxy

The PowerXXXX design system (themes, colours, logo, layout) is shared with PowerSolver: see PowerSolver's documentation/LAYOUT_GUIDE.md.

The product in one picture

Excel workbook / CSV  ──►  Wizard (4 steps)  ──►  Forecast engine  ──►  Results page
history + settings         Goal · Your data       checks each series      verdict, chart, table,
                           How far ahead ·        tries several methods   your data, quality checks,
                           Review                 on your past, keeps     AI critic
                                                  the most accurate

Where the files are

Folder	What
`forecast-wizard/`	the wizard and Results page (static HTML/JS)
`forecast-api/`	the forecast engine (Python, FastAPI, StatsForecast)
`forecast-demos/`	the demos (each with its `input/*.xlsx`)
`forecast-datasets/`	the real public datasets the demos and tests use, with their provenance (`MANIFEST.md`)
`forecast-templates/`	real-data templates used by the engine tests
`Documentation/`	this documentation; `python -I Documentation/tools/build_html.py` builds the HTML site