Penalized Estimation in Large-Scale Generalized Linear Array Models

Large-scale generalized linear array models (GLAMs) can be challenging to fit. Computation and storage of its tensor product design matrix can be impossible due to time and memory constraints, and previously considered design matrix free algorithms do not scale well with the dimension of the parameter vector. A new design matrix free algorithm is proposed for computing the penalized maximum likelihood estimate for GLAMs, which, in particular, handles nondifferentiable penalty functions. The proposed algorithm is implemented and available via the R package glamlasso. It combines several ideas-previously considered separately-to obtain sparse estimates while at the same time efficiently exploiting the GLAM structure. In this article, the convergence of the algorithm is treated and the performance of its implementation is investigated and compared to that of glmnet on simulated as well as real data. It is shown that the computation time for glamlasso scales favorably with the size of the problem when compared to glmnet. Supplementary materials, in the form of R code, data and visualizations of results, are available online.