AbstractProjection pursuit is a technique for finding highly informative low‐dimensional projections of multivariate data for visual inspection by an analyst. When data of dimension m are reduced to dimension p (where typically p = 2 is the most useful for viewing by scatter plots), the method consists of defining a measure of information content in two dimension and optimizing that measure or index as a function of two m‐dimensional projection vectors to find the most informative projection. The main ideas that have guided research in projection pursuit are reviewed with the emphasis on ideas leading to practical implementations. The most common indices investigated and used are described, and comparative studies of these indices as used in projection pursuit are reviewed. Other aspects of implementation such as the choice of optimization algorithm are also covered. Copyright © 2009 Wiley Periodicals, Inc., A Wiley CompanyThis article is categorized under: Statistical and Graphical Methods of Data Analysis > Analysis of High Dimensional Data