RAGNU: A microcomputer package for two-group mathematical programming-based nonparametric classification

In this manuscript, we introduce the PC-based software package RAGNU, a utility program that can be used, in conjunction with the LINDO optimization software, for solving two-group classification problems using a class of recently developed nonparametric methods. The criteria used to estimate the classification function are based on either minimizing a function of the absolute deviations from the surface that separates the groups, or directly minimizing a function of the number of misclassified observations. Since mathematical programming (MP) techniques are efficient tools for analyzing such problems, we will refer to this class of nonparametric methods as MP-based methods. Recently, a number of research studies have reported that under certain data conditions MP-based methods can provide more accurate classification results than existing parametric statistical methods, such as Fisher's linear discriminant function and logistic regression. It has also been shown that extensions of the MP-based formulations that incorporate non-linear (e.g., quadratic) functions of the attribute values are a viable alternative to Smith's quadratic discriminant function. However, these robust MP-based classification methods have not yet been implemented in the major statistical packages, and hence are beyond the reach of those statistical analysts who are unfamiliar with mathematical programming techniques. Currently, only those researchers who have written their own interface software programs are able to use MP-based classification methods. Therefore, we believe that RAGNU contributes significantly to the field of nonparametric classification analysis, in that it provides the research community with convenient access to this class of robust methods. RAGNU is available from the authors without charge.