|
Abstract:
|
The purpose of this talk is to introduce a new classification methodology. The methodology uses a connection, which we uncover, between classification and testing, and is called Test-Based Classification Rule. In particular, when a rank-based testing method is used, we call it Rank Test-Based Classification. We start from binary classification in terms of univariate and multivariate data, and then move on to the multiclass case. Several simulated data sets and real data sets are covered in this talk. We argue that our new idea is competitive with the Linear and Quadratic Classification Rules when the observed data are normally distributed, but it can be a better alternative than the Linear and Quadratic Rules in most of cases when data are not normal. Lanchenbruch's holdout misclassification rate is used as a tool to judge the performance of classification.
|
