Generalised additive dependency inflated models including aggregated covariates (Q1711602)

The ``Generalised Additive Dependency Inflated Model including Aggregated Covariance'' (GADIMAC) model has the form \[ U = G(m_0 + m_1(X) + m_2(\lambda(X)Y) + m_3(X+Y)) + \epsilon, \] where $X, Y, U$ are observable, $G$ is a known invertible link function, \(m_0\) is an unknown constant, and $m_1, m_2, m_3, \lambda$ are unknown functions. The paper under review presents results on the rate of a penalized least squares estimator of $(m_0, m_1, m_2, m_3, \lambda)$ and on the identification of those functions. Furthermore, an age-period-cohort density version of GADIMAC is applied to forecast future asbestos-related deaths in the UK.