CMStatistics 2017: Start Registration
View Submission - CMStatistics
Title: fsdaR: making the FSDA toolbox available to R users Authors:  Emmanuele Sordini - Joint Research Centre of the European Commission (JRC) (Italy) [presenting]
Valentin Todorov - UNIDO (Austria)
Aldo Corbellini - Faculty of Economics - University of Parma (Italy)
Abstract: The FSDA (Flexible Statistics for Data Analysis) toolbox is a software library that extends MATLAB and its Statistics Toolbox to support a robust and efficient analysis of complex datasets, affected by different sources of heterogeneity. FSDA was born around the forward search approach, and has evolved into a comprehensive and computationally efficient software package for robust statistics. FSDA provides tools for regression and multivariate analysis, robust transformations, cluster analysis and model selection. FSDA provides a rich set of graphical features not available in R, such as dynamic brushing, linking, etc., particularly useful for exploratory data analysis. The only downside is that FSDA is based on commercial software (MATLAB), which is not so appealing to the majority of the statistical community, where R is more widespread. Therefore, in 2016 we developed a prototype in which we demonstrated the feasibility of calling MATLAB functions in R in order to make selected features of FSDA available to R developers. The first prototype has since evolved into a package (named fsdaR) compliant with R coding and calling standards (object-oriented approach, formula notation, etc.), providing to R users selected features of the FSDA toolbox along with some of its advanced graphical capabilities. We will focus on i) The structure of the fsdaR package; ii) computational and graphical features of FSDA available in R through fsdaR, iii) A live demo.