An R Package for Nonparametric Likelihood Ratio Tests for Goodness-of-Fit and Two-Sample Comparisons Based on Sample Entropy

• Activity Number: 32
• Type: Contributed
• Date/Time: Sunday, July 31, 2011 : 2:00 PM to 3:50 PM
• Sponsor: Section on Statistical Computing
• Author(s): Jeffrey Miecznikowski*+ and Albert Vexler and Lori Shepherd
• Companies: The State University of New York at Buffalo and New York State University at Buffalo and The State University of New York at Buffalo
• Keywords: empirical likelihood ; likelihood ratio ; goodness-of-fit ; sample entropy ; nonparametric tests ; two-sample comparisons

Abstract

We introduce and examine db.emp.like.gof, an R language for performing goodness-of-fit tests based on sample entropy. This package also performs the two sample distribution comparison test. For a given vector of data, the provided functions test the data for the proposed null distributions, or test for distribution equality between two vectors of observations. The proposed methods represent a distribution-free density-based empirical likelihood (EL) technique applied to nonparametric testing. The proposed procedure performs exact and very efficient p-values for each test statistic obtained from a Monte-Carlo (MC) resampling scheme. Note by using an MC scheme, we are assured exact level alpha tests that approximate nonparametrically most powerful Neyman-Pearson decision rules. Although these entropy based tests are known in the theoretical literature to be very efficient, they have not be