Percolation and multimodal data structuring

  • Raymond C. Trémolières
Conference paper
Part of the Studies in Classification, Data Analysis, and Knowledge Organization book series (STUDIES CLASS)


In this article the problem of multimodal data analysis is considered in the framework of the percolation method. A new presentation is given and we summarize our experience. It must be recalled that it is the only method in data analysis to permit a classification of points into four categories: group points, boundary points, multimodal points and isolated points. These concepts allow to correctly describe the shape of any empirical density functions. Points related to same modes are ’group points’. The others permit to describe the relative position of groups with respect to themselves. So the method appears to be a hybrid one between clustering analysis and pattern recognition. Comparisons with other data analysis methods are given.


Boundary Point Group Point Support Point Dissimilarity Matrix Perception Level 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. MAHFOUDI, A., TREMOLIERES, R. (1990): How to compute correlation coefficients in qualitative and quantitative data analysis. Actes des 5èmes Journées Internationales des Sciences Informatiques, Tunis, 9–11 mai 1990.Google Scholar
  2. TREMOLIERES, R. (1979): The percolation method for an efficient grouping of data. Pattern Recognition, 11, 4. CrossRefGoogle Scholar
  3. TREMOLIERES, R. (1984): The generalized percolation for data analysis and pattern recognition. In: J. Janssen et al. (eds.): New Trends in Data Analysis and Application. North-Holland.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1994

Authors and Affiliations

  • Raymond C. Trémolières
    • 1
  1. 1.IAEUniversity of Aix-Marseille IIIPuyricardFrance

Personalised recommendations