RINK DocsForecast smarter, faster.

Time-series forecasting on your data — upload, train, and predict in one workspace.

📊

Auto-detected schemas

Date and numeric columns are detected automatically. Reverse-chronological CSVs are sorted to ascending order before training.

🧠

Lag features (1, 2, 3, 5, 7) and rolling means (3, 7) feed a Gradient Boosting model with a held-out validation split.

🔮

Recursive forecasting up to 200 steps with a confidence band that widens with the horizon.

🔐

Each user's CSV and model live under their own UUID on the server, and are deleted automatically on sign-out.

⏱

Sessions auto-terminate after four hours of inactivity. Activity is tracked across tabs.

🤖

Ask the in-app assistant about ML, forecasting techniques, or how to use the platform.

Getting Started — sign up, upload a dataset, generate your first forecast in five minutes.
Architecture — how the React app, Express gateway, and FastAPI ML service fit together.
Guides — task-by-task walkthroughs for everything you can do in the workspace.
API Reference — complete request/response docs for every endpoint.
Deployment — push your own copy to Render + Vercel and onto a custom domain.
FAQ — answers to the questions everyone asks first.

New here? Start with Getting Started. It will walk you through your first upload, training run, and forecast.