Identify influential points
Lin, H.; Su, Y.
Wei Sheng Yan Jiu 27(5): 355-357
1998
ISSN/ISBN: 1000-8020 PMID: 11939058 Document Number: 496461
Influential points are the points with excessive influence on result and detected by influential function theta-theta(i). theta-theta(i) is a multidimensional vector. For convenience, people always consider some quantitative function g (theta-theta(i)) to measure the vector theta-theta(i), but how to define g is a difficult problem. This paper proposes a new quantitative diagnostic statistics for identifying influential points for all parametrical statistical analysis. The method is more appropriate than cook's method for regression analysis. In this study an example is used to illustrate the procedure in COX proportional hazard model.