Projection Pursuit
Abstract
Projection pursuit (PP), which goes back to Kruskal and Friedman and Tukey, is discussed from a conceptual point of view. Originally, it was merely concerned with finding 'interesting' projections to aid the visual analysis of high dimensional data. It will be argued that there is a very general concept behind PP, having a much broader scope than anticipated by its originators. In its most important aspect it is concerned wind finding least normal projections of the data, and it has ramifications reaching into topics as varied as robust covariance estimation, factor analysis, nonparametric signal detection, computer tomography and Hilbert's 13th problem.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 1981
- Accession Number
- ADA107817
Entities
People
- Peter J. Huber
Organizations
- Harvard University