A0277
Title: Visualising model selection stability
Authors: Garth Tarr - University of Sydney (Australia) [presenting]
Abstract: Various proposals to understand the stability of selected features from various models have been proposed over the past 15 years. Some of these are investigated from the lens of how to communicate selection stability results in a visual way to aid understanding and interpretation for statistical practitioners. A key tool is the mplot R package which has the aim of helping researchers implement stability selection concepts to use to better inform the variable selection process. The focus is on recent improvements to the mplot package that builds on recent advances in exhaustive search techniques and on separating out stably selected features in regularized regression models.
