Fig. 3
From: ecpc: an R-package for generic co-data models for high-dimensional prediction
Illustration of the non-linear co-data models. Given the true effect sizes \(\beta _k^2\) (points), a linear co-data model may capture the relation insufficiently, making estimation of non-linear relations desirable. A generalised additive model estimates a smooth non-linear relation. Additionally, one may further impose constraints, e.g. monotonicity (left) or convexity (right), in a shape-constrained additive model