The Horseshoe-Like Regularization for Feature Subset Selection

The design and analysis of non-randomized studies: a case study of off-label use of hydroxychloroquine in the COVID-19 pandemic

Introduction: The COVID-19 pandemic has prompted researchers to conduct non-randomized studies in an effort to find an off-label drug that can effectively combat the virus and its effects. While these studies can expedite the drug approval process, researchers must carefully design and analyze such studies in order to perform rigorous science that is reproducible and credible. This article focuses on several key design and analysis considerations that can improve the scientific rigor of non-randomized studies of off-label drugs. Areas covered: The aim of this article is to provide an overview of best approaches that should be considered for non-randomized studies on off-label drugs. We discuss these approaches in detail and use a non-randomized study by Rivera et al. in Cancer Discovery as an example of methods that have been undertaken for COVID-19. Expert opinion: While non-randomized studies are inherently biased, they may be unavoidable in situations such as the COVID-19 pandemic, where researchers need to find an effective treatment quickly. We believe that a well-formed experimental design, high-quality data collection, and a well-thought-out statistical and data analysis plan are sufficient to produce rigorous and credible results for making an optimal decision.

Triple the Gamma—A Unifying Shrinkage Prior for Variance and Variable Selection in Sparse State Space and TVP Models

Time-varying parameter (TVP) models are very flexible in capturing gradual changes in the effect of explanatory variables on the outcome variable. However, in particular when the number of explanatory variables is large, there is a known risk of overfitting and poor predictive performance, since the effect of some explanatory variables is constant over time. We propose a new prior for variance shrinkage in TVP models, called triple gamma. The triple gamma prior encompasses a number of priors that have been suggested previously, such as the Bayesian Lasso, the double gamma prior and the Horseshoe prior. We present the desirable properties of such a prior and its relationship to Bayesian Model Averaging for variance selection. The features of the triple gamma prior are then illustrated in the context of time varying parameter vector autoregressive models, both for simulated dataset and for a series of macroeconomics variables in the Euro Area.