TY - RPRT T1 - Applying permutation tests to tree-based statistical models: extending the R package rpart Y1 - 2004 A1 - Cummings, Michael P. A1 - Myers,D. S A1 - Mangelson,M AB - Tree-based statistical models are useful for evaluating relationships between predictor and response variables and for generating predictions when the response is unknown. However, current methods of constructing tree-based models do not provide a probabilistic assessment of the models produced. Here we describe our work to use permutation tests to quantitatively estimate the probability of tree-based statistical models. We have extended the rpart (recursive partitioning) package of the R system for statistical data analysis. This extension, rpart.permutation, executes the permutations in parallel, using MPI (Message Passing Interface) to greatly decrease the time necessary to complete the analysis. PB - University of Maryland Institute for Advanced Computer Studies VL - 2004-24 ER -