Testing Exponentiality Against UBAC Using Kernel Methods

Abstract

In this paper, the problem of testing exponentiality against used better than aged in convex ordering classes of life distributions is investigated. For this property a nonparametric test is presented based on kernel method. The test is presented for complete and right censored data. Furthermore, Pitman's asymptotic relative efficiency (PARE) is discussed to assess the performance of the test with respect to other tests. Selected critical values are tabulated. Some numerical simulations on the power estimates are presented for proposed test. Finally, examples in medical sciences are used as practical applications for the proposed test.

Keywords

• UBAC Classes of Life Distributions; Kernel Method; Survival function; Right censored data; Kaplan-Meier estimator