| 讲座简介： | Although model selection is ubiquitous in scientific discovery, the stability and uncertainty of the selected model is often hard to evaluate. To this goal, the concept of model confidence bounds (MCB) in the context of nested model is introduced. MCB identifies two nested models (upper and lower confidence bound models) containing the true model at a given level of confidence, whose width and composition can be the measure of the model selection uncertainty. Several graphical tools are further proposed. These include the model uncertainty curve (MUC), G-plots, H-plots, and model selection deviation (MSD), to visualize the distribution of the selected model and the variability of model selection. The proposed assessment method and visualization tools enable the practitioner to evaluate the overall model selection uncertainty and to compare different model selection procedures. Monte Carlo simulations and the real data examples confirm the validity and illustrate the advantages of the proposed method. |
