PIONEER: Distance-based multivariate analysis

PIONEER is a program for the analysis of (social science) data of the form subjects (respondents, objects, etc.) by variables. It emphasizes a graphical representation of the subjects. The program is particularly suited to explore how similar the subjects are. PIONEER is the acronym of Program for Individual-Oriented Numeration Embedded in Empirical Research.

This page contains information about the following topics:

Example of PIONEER data

Consider, as an example, an exploratory medical study for diagnosing the severity of an illness, where several diagnostic variables are measured. These variables can be used to find a representation that summarizes the differences between the patients. This representation can also show how these differences are related to the severity of the illness. Therefore, an accurate representation of the patients is quite important. PIONEER emphasizes the graphical representation of subjects, such that similar subjects are represented close together and dissimilar subjects far apart.

Features of PIONEER

  • Emphasis on geometrical representation (a map) of subjects; subjects that are close together are similar, those far apart are very different.
  • Variables can be optimally transformed: nominal (categorical) variables, ordinal variables, numerical, spline transformation.
  • Subjects can be weighted differentially, for example by their frequency in the sample.
  • Missing data can be handeled.

The PIONEER program

  • PIONEER can be downloaded freely from this page.
  • PIONEER has a simple SPSS-like command interface.
  • All plots by PIONEER are printed as a line plot in the output. High resolution versions of these plots are saved in a postscript file. The Macintosh version has in addition high resolution plots available on screen.
  • Current operating systems supported: Windows 95, MacOS, Sun Solaris.

References

Below you will find PIONEER-related references.

  • Meulman, J.J. (1986). A distance approach to nonlinear multivariate analysis. Leiden: DSWO Press, Leiden University.
  • Meulman, J.J. (1992). The integration of multidimensional scaling and multivariate analysis with optimal transformations. Psychometrika, 57, 539-565.
  • Groenen, P.J.F., Commandeur, J.J., and Meulman, J.J. (1997). PIONEER: a program for distance-based multivariate analysis. In: Proceedings of SoftStat 97.

Downloading PIONEER

You can now download (click to download) version 0.31 of PIONEER and the user's guide. PIONEER is available for the following operating systems. We would like to hear any remarks about the program from you.

Bart Jan van Os (os@fsw.leidenuniv.nl)

Page last modified on Thursday, April 01, 2004.