|
Activity Number:
|
72
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Sunday, August 2, 2009 : 4:00 PM to 5:50 PM
|
|
Sponsor:
|
Section on Statistical Computing
|
| Abstract - #303434 |
|
Title:
|
Wrapped-Around KS Type Statistics
|
|
Author(s):
|
Huiyu Qian*+ and Wei-Min Huang and Bennett Eisenberg
|
|
Companies:
|
Lehigh University and Lehigh University and Lehigh University
|
|
Address:
|
14 E Packer Ave. , Bethlehem, PA, 18015,
|
|
Keywords:
|
Kolmogorov-Smirnov ; Directional data ; Wrapped-around ; Goodness-of-fit ; Kuiper's test ; Data-driven
|
|
Abstract:
|
The Kolmogorov-Smirnov (KS) test has its advantage of being nonparametric and distribution-free with the cost of having relatively low power generally. Moreover, it can not be applied to directional data. The proposed wrapped KS type statistic applies to both data on a line and on a circle. The statistic is defined as the maximum epsilon-interval distance between the empirical and hypothesized distributions in terms of weighted CDFs. A special case of the statistic is equivalent to the Kuiper's test statistic. The wrapped KS type tests are compared with some other well-known goodness-of-fit tests. Asymptotic properties and Mante Carlo simulations are included in the studies.
|
