Google Research has released TimesFM-3, an AI model that forecasts the future from time series like daily sales figures, drawing on related data and known upcoming events to improve its predictions.
Real-world forecasts rarely depend on a single variable, and Google illustrates this with a retail chain trying to predict ice cream sales. A good forecast should also factor in related products like waffle cones or syrup, along with past foot traffic, weather, discount campaigns, and holidays.
TimesFM-3 is built on a Transformer, the same base architecture as its predecessors, but it groups 32 consecutive data points into a single patch and normalizes each series to a common scale so measurements of very different magnitudes can be compared directly.
The model processes data in two alternating directions. Along the time axis, it looks for patterns within a single series, only drawing on past values to avoid leaking future information. Across series, it compares all variables at a given point in time and learns how they relate, which lets it pick up on things like how a discount on one product affects sales of another.
Blue marks the target series, purple a variable known only historically, and green a known future event whose tokens already include upcoming patches. | Image: Google
