Skip to content
ML Academy

Introduced on day 27

Linear check

A simple linear forecaster over recent history, trained across series, used to measure how much a more complex model adds beyond a linear map.

In detail

Inspired by results where one-layer linear models matched or beat several Transformer forecasters on long-horizon benchmarks. It sits between the seasonal baseline and a heavy family: if it recovers most of the available gain, the remaining headroom for a complex model is small.