Title: MultiKurt: An R package for kurtosis-based data analysis
Authors: Cinzia Franceschini - Tuscia University (Italy) [presenting]
Nicola Loperfido - University of Urbino (Italy)
Abstract: Multivariate kurtosis plays an important role in several areas of multivariate analysis: normality testing, projection pursuit, independent component analysis, model-based clustering, portfolio optimization and density approximation. MultiKurt is an R package purported to describe, testing and visualize multivariate kurtosis. In particular, it incorporates state-of-the-art algorithms for computing linear projections which either maximize, minimize or remove kurtosis. The package also computes scalar-valued and matrix-valued measures of multivariate kurtosis.