Skewness Projection Pursuit: Past, Present and Future

Skewness projection pursuit is a multivariate statistical method which looks for interesting data projections via the optimization of their third absolute cumulants. This paper reviews the literature on skewness projection pursuit, including theoretical foundations, computational aspects, critical issues and future research. Special emphasis is given to the connection between skewness projection pursuit and the canonical form of the skew-normal distribution and its generalizations. Model-based and data-based examples illustrate the concepts.