Abstract:
For numerous applications it is of interest to provide full probabilistic forecasts, which are able to assign probabilities to each predicted outcome. Therefore, attention is shifting constantly from conditional mean models to probabilistic distributional models capturing location, scale, shape (and other aspects) of the response distribution. One of the most established models for distributional regression is the generalized additive model for location, scale and shape (GAMLSS). In high dimensional data set-ups classical fitting procedures for the GAMLSS often become rather unstable and methods for variable selection are desirable. Therefore, we propose a regularization approach for high dimensional data set-ups in the framework for GAMLSS. It is designed for linear covariate effects and is based on L1-type penalties. The following three penalization options are provided: the conventional least absolute shrinkage and selection operator (LASSO) for metric covariates, and both group and fused LASSO for categorical predictors. The methods are investigated both for simulated data and for two real data examples, namely Munich rent data and data on extreme operational losses from the Italian bank UniCredit.