Bootstrapping to assess and improve atmospheric prediction models

Bootstrapping is a simple technique typically used to assess accuracy of es timates of model parameters by using simple plug-in principles and replacin g sometimes unwieldy theory by computer simulation. Common uses include var iance estimation and confidence interval construction of model parameters. It also provides a way to estimate prediction accuracy of continuous and cl ass-valued outcomes regression models. In this paper we will overview some of these applications of the bootstrap focusing on bootstrap estimates of p rediction error, and also explore how the bootstrap can be used to improve prediction accuracy of unstable models like tree-structured classifiers thr ough aggregation. The improvements can typically be attributed to variance reduction in the classical regression setting and more generally a smoothin g of decision boundaries for the classification setting. These advancements have important implications in the way that atmospheric prediction models can be improved, and illustrations of this will be shown. For class-valued outcomes, an interesting graphic known as the CAT scan can be constructed t o help understand the aggregated decision boundary. This will be illustrate d using simulated data.