Flexible Multidimensional Scaling with the R Package smacofx

Thomas Rusch, Patrick Mair, Jan de Leeuw, Kurt Hornik

Main Article Content

Abstract

In this article we introduce the R package smacofx which offers various implementations of flexible multidimensional scaling (MDS) techniques for two-way data. These techniques allow for flexible MDS via utilizing combinations of optimal scaling and specific parametric transformations of the input dissimilarities and/or the fitted distances, and/or residual weighting. Examples of techniques available in the package include Sammon mapping, power-stress and r-stress MDS, Box-Cox MDS, local MDS, (extended) curvilinear component analysis and curvilinear distance analysis. Various post-fit functionality is available that matches the functionality available in smacof (De Leeuw and Mair 2009), including a broad array of plots to visualize and communicate the results. The package further includes tools and functions for goodness-of-fit assessment, local optimum diagnostics and MDS biplots. All the package functionalities are illustrated using two data applications.

Article Details

Article Sidebar