Multivariate Robust Statistics

The goal of robust statistics is to develop methods that can cope with the presence of outliers in the data and nevertheless produce reasonable results. In this book some of the most popular robust multivariate methods are investigated and new methods are proposed. Their performance is evaluated and compared in a variety of situations. The focus is on high breakdown point methods for discriminant analysis, multivariate tests and their basis, the robust estimators for multivariate location and covariance. The routine use of robust methods in a wide area of application domains is unthinkable without the computational power of today's personal computers and the availability of ready to use implementations of the algorithms. A unified computational platform organized as common patterns which we call statistical design patterns in analogy to the design patterns widely used in software engineering is proposed. The concrete implementation is an object oriented framework for robust multivariate analysis developed in R, an environment for statistical computing and graphics (R Development Core Team, 2009).

Autorentext

Valentin Todorov is a Management Information Officer with theUnited Nations Industrial Development Organization (UNIDO). Hereceived a doctoral degree from Vienna University of Technology,Austria.Peter Filzmoser is a professor at Vienna University ofTechnology. He holds a doctoral degree and habilitation from thesame University.