Geoff Negiar

And while our own platform only runs foundation models, we are working on allowing you to upload your forecasts, however you've obtained them. That way you can compare any method to foundation models, and go into the weeds of what's accurate or not.

The Forecasting Company A new foundation model that can predict any time series

Apply for access on our website www.theforecastingcompany.com

Evaluation is super important in time series forecasting. This is why we're building the first interactive benchmarking and evaluation platform at @theforecastingco.bsky.social.

We believe precise and granular evaluation trumps single top-level metrics for actual industry use-cases.

Thanks for the welcome! We're not focusing on physics heavy systems like weather. There's a lot of very domain specific things you can/need to do to be competitive in weather, e.g. integrating physics into the model. Our friends at @atmo-ai.bsky.social, Vayuh.ai and Silurian.ai would be the ones!

Thank you so much Eugene, both for being with us and for the shout out 🙏

I've been very excited about Geoff's recent pursuit of a foundation model for time-series data. If you're interested in that goal, definitely check him (and his company!) out

Hello 🦋!
This is my first post on this platform -- thank you all for following me. We're building the future of time series systems - starting with simple numerical models + benchmarking. Follow along if you're interested in forecasting systems for your business, or in modern times series research!

Keith jarrett tho