BinCom (Mazanec, 2008) analyzes binary data. The pattern counting procedure PATCLU transforms the raw data vectors into a smaller data set for subsequent clustering. PATCLU determines the absolute frequencies of each data vector and prepares the input for the BINCLUS routine. BINCLUS then operates on the response patterns weighted by their frequency of occurrence. Depending on the number of repeated occurrence of the individual response data the volume of input data gets reduced. In the demo data set, e.g., 3038 individual records of 'pick-up-to-three-from-20' zero-one data are compressed into 549 patterns. This reduction accelerates the processing with the exchange algorithm implemented in BINCLUS.

BinCom is based on the binary clustering procedure by Späth, H. (1977), Partitionierende Cluster-Analyse für große Objektmengen mit binären Merkmalen am Beispiel von Firmen und deren Berufsgruppenbedarf, in: Späth, H. (ed.), Fallstudien Cluster-Analyse, München-Wien, pp. 63-80 and on the pattern counting mechanism introduced in Formann, A. K., J. A. Mazanec, and O. C. Oberhauser (1979), Numerische Klassifikationsprobleme in 'großen' Datensätzen der demoskopischen Marktforschung: Ein empirischer Methodenvergleich von Latent Class- und Cluster-Analyse, Arbeitspapiere der absatzwirtschaftlichen Institute der Wirtschaftsuniversität Wien, Nr. 12/1979. An outline is given in Mazanec, J. A. (1997), Advanced methodology for detecting competitors, mapping destinations, constructing tourist types and analysing competitive positions, in: International City Tourism: Analysis and Strategy, ed. by J. A. Mazanec et al., London-Washington: Pinter, pp. 242-245.

The PATCLU and BINCLUS programs are new Watcom-compiled versions. For facilitating usage the BinCom commander assists in parameterizing the tandem procedures.

Note: Please read and accept our download conditions before!