| Biography | |
|---|---|
 Prof. Jin Wang Northern Arizona University, USA |
|
| Title: Effect of Dimension and Kurtosis on Performances of Some Algorithms | |
| Abstract: I modern statistics, almost all statistical methods are implemented through algorithms. Thus performance of a statistical method is directly affected by the algorithm for the method. Here we study two well-known algorithms for multivariate data. It is found that the performances of both algorithms decline as dimension increases. The effect of data shape on the algorithms is also studied. Our finding is that the performances of both algorithms decrease as kurtosis increases. Some adjustments for those algorithms will be discussed as well, along with some new descriptive measures for multivariate data. | |
| Biography: Jin Wang received his Ph.D. in statistics from the University of Texas at Dallas and joined Northern Arizona University (NAU) in 2003. His research on statistics started from the multivariate change-point problems at Wuhan University in 1990. The paper was invited to present at the International Symposium on Multivariate Analysis and Its Applications, Hong Kong, in 1992, and published in the IMS Lecture Notes – Monograph Series. His recent research focused on nonparametric multivariate analysis and its applications. The following are some representative works. Wang and Serfling (2005) introduced a nonparametric multivariate kurtosis measure. The measure is not only robust but also discriminates better among distribution shapes. It determines elliptically symmetric distributions up to affine equivalence. In 2009, he proposed a family of kurtosis orderings for multivariate distributions, which is a pioneering work on multivariate kurtosis ordering. Various applications of the orderings have appeared in the literature. Furthermore Wang and Zhou (2012) defined a generalized multivariate kurtosis ordering. Based on the ordering, they developed a two-dimensional visual device to compare two distributions in any dimension with respect to spread and kurtosis. In a publication this year, he studied the asymptotic behavior of generalized depth-based spread processes. Based on the results, he designed a graphical method to compare spread and kurtosis of two multivariate data sets and a new graphical method to assess multivariate normality. Besides the theoretical researches, Dr. Wang is also interested in applications of statistics in various fields to solve practical problems. He worked in industry for eight years (1991-1999) and currently participated in several health-related projects at NAU. | |
