主 講 人：黃郁芬 教授 中正大學 數學系
講    題：Sensitivity Analysis of Nongaussianity by Projection Pursuit 
日    期：99年12月28日（星期二）下午2:30 – 3:20 
地    點：數學系(科學館S433室) 
摘    要： 
From the information-theoretic point of view, the Gaussian distribution is the 
least structured. Therefore, the most non-Gaussian direction in which to explore 
the clustering structure of data is considered to be the most interest-ing 
projection direction when applying projection pursuit. Non-Gaussianity is often 
measured by kurtosis. However, kurtosis is well-known to be sensitive to 
in.uential points/outliers and so the projection direction can be unduly a.ected 
by abnormal points. In this paper, we focus on developing in.uence functions of 
projection directions in order to detect abnormal observations, especially on 
high-dimensional data. For multivariate data, a new technique is proposed for 
de.ning and developing in.uence functions of projection di-rections. In addition, 
a new in.uence function is suggested. Two simulated data examples and one 
concrete data example are provided for illustration. 

Keywords: In.uence Function, Kurtosis, Non-Gaussianity, Projection Pur-suit.