An Intelligent Clustering Technique Based on Dual Scaling
Methods of cluster analysis (unsupervised classification) can help you in order to “Finding groups in data”, so the suitable title of a book from Kaufman and Rousseeuw (1990). The intelligent clustering technique proposed here appears to be motivated by practical problems of analyzing mixed data. One way to deal with such problems is downgrading all data to the lowest scale level, that is, downgrading to categories without any information about their order. This new clustering technique based on dual scaling is presented in the simplest fashion possible for finding two groups (clusters) in data and for visualizing them, respectively. It is compared with well-known model-based clustering techniques. In the conclusion variations of improvement of the method are proposed.
KeywordsMisclassification Error Principal Component Analysis Plot Hierarchical Cluster Method Distance Score Time Standard Deviation
Unable to display preview. Download preview PDF.
- Breiman, L. (1996). Bias, variance, and arcing classifiers. Technical Report 460. Statistical Department, University of California, Berkeley.Google Scholar
- Fraley C. (1996). Algorithms for model-based Gaussian Hierarchical Clustering. Technical Report 311. Department of Statistics, University of Washington, Seattle.Google Scholar
- Hand, D. J. (1997). Intelligent data analysis: issues and opportunities. In: Advances in Intelligent Data Analysis. Liu, X. et al. (eds.), 1–14, Springer-Verlag, Heidelberg.Google Scholar
- Mucha, H.-J. (1992). Clusteranaly_se mit Mikrocomputern. Akademie Verlag, Berlin.Google Scholar
- Mucha, H.-J., Siegmund-Schultze, R., and Dübon, K. (1998). Adaptive cluster analysis techniques–software and applications. In: Data Science, Classification and Related Methods, Mucha, H.-J., Siegmund-Schultze, R., and Dübon, K. (eds.), 231–238, Springer-Verlag Tokyo.Google Scholar
- Nishisato, S. (1994). Elements of Dual Scaling: An Introduction to Practical Data Analysis. Lawrence Erlbaum Associates, Publishers, Hillsdale.Google Scholar