What is Projection Pursuit?

From MaRDI portal





The paper gives an overview of projection pursuit problems. It explains the background, connections with other multivariate data processing methods and relations of the present authors' work with other ones. Types of projection indexes (Friedman-Tuckey, entropy, moment, or cumulant) for one-dimensional and two-dimensional projections as well as centering and sphering data are discussed. Also, implementation problems as selecting of window width, methods for finding minimum (maximum) of indexes, starting points, stopping rules etc. are treated. An example using the Lubischew data (74 cases of 6 measurements on male fleabeetles of three species) proves (for one-dimensional projections as well as for two-dimensional) the power of the method. The paper is finished with a description of projection pursuit density estimation and regression analysis. Interesting discussion-contributions of some well- known statisticians conclude this paper which expresses a little different point of view from that one of the analogous Huber's paper and it makes it even more interesting.




Cited in
(only showing first 100 items - show all)








This page was built for publication: What is Projection Pursuit?

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3769796)