GINKGO, A PROGRAM FOR NON-STANDARD MULTIVARIATE FUZZY ANALYSIS
Many multivariate programs are expensive commercial packages or require expensive third party software. Other applications are freely available to academic researchers but are limited to one operating system. This paper presents GINKGO, a free easy-to-use multi-platform Java application mainly oriented towards to non-standard classification analysis. This orientation is pursued (1) by providing users with several similarity and dissimilarity measures, (2) by allowing the execution of several crisp or fuzzy prototype-based clustering methods on arbitrary distance matrices, and (3) by including standard and non-standard facilities for evaluating the quality of clusters. Along with the software presentation, two methodological improvements are also given, which are available in the program. First, the implementation of a new parameter initialization strategy for possibilistic C-means. As an example to illustrate both the methodological advance and the capabilities of the program, a clustering analysis of human fibroblast DNA microarray expression data is presented. Second, a fuzzy generalization of the Rand and corrected Rand indices to allow the comparison of fuzzy partition matrices. GINKGO is available at the website http://biodiver.bio.ub.es/ginkgo, and software updates are automatically done via Java Web Start technology.
classification, multivariate data representation, fuzzy clustering analysis, statistical software.