Contents

• PREFACE
• LIST OF PUBLICATIONS
• THE AUTHOR'S CONTRIBUTION
• LIST OF SYMBOLS AND ABBREVIATIONS
• INTRODUCTION
• METHODS FOR EXPLORATORY DATA ANALYSIS
  • Visualization of high-dimensional data items
  • Clustering methods
  • Projection methods
    • Linear projection methods
      • Principal component analysis
      • Projection pursuit.
    • Nonlinear projection methods
      • Multidimensional scaling
      • Principal curves.
      • Other methods.
  • Self-organizing maps
    • The self-organizing map algorithm
    • Properties useful in exploring data
      • Ordered display.
      • Visualization of clusters.
      • Missing data.
      • Outliers.
    • Mathematical characterizations
      • Relation to K-means clustering.
      • Relation to principal curves.
      • A decomposition of the cost function.
    • Some variants
    • Notes on statistical accuracy
  • Relations and differences between SOM and MDS
    • Relations of the mappings in the ideal case.
    • Computational properties.
    • Combinations of the methods.
• STAGES OF SOM-BASED EXPLORATORY DATA
  ANALYSIS
  • Preprocessing
  • Computation of the maps
  • Choosing good maps
    • Note 1: Sparse data.
    • Note 2: High dimensionality.
    • Note 3: Computational complexity.
  • Interpretation, evaluation, and use of the maps
    • Interpretation.
    • Evaluation.
    • Use of the organized maps for exploratory data analysis.
• CASE STUDIES
  • Multichannel EEG signal
  • Statistical tables
  • Full-text document collections
    • Recent developments
• FURTHER DEVELOPMENTS
  • Feature exploration with the adaptive-subspace SOM
  • Comparison of knowledge areas
• CONCLUSION
• References
• APPENDIX: KEY TO THE COUNTRY NAMES
• About this document ...