Numerical Cruncher

User's Guide

Menu Tree

• Data (i.e. the typical "File" Menu)
  • Statistics to get some information about input data
  • 2D Plot to view pairs of attribute values
  • Exit. As it name implies...
  
  
• Edit is used to modify the training set
  (e.g. to reduce its size before building a k-NN classifier)
  • Undo edition restores the original training set
  • Wilson's edition
  • Partition edition
  • Multiedit
  • Hart's condensation
  
  
• Classification collects some algorithms widely used to build classifiers
  • Classes... shows class stats
  • Parametric models
    • Euclidean Distance
    • Normalized Euclidean Distance
    • Mahalanobis' Distance
    • Quadratic Classifier
    • Parallelepipeds Method
  • Nearest Neighbour(s)
    • 1-NN
    • k-NN
  • Adaptive Learning algorithms
    • LVQ-1 [Kohonen's Learning Vector Quantization]
    • OLVQ-1, Optimized LVQ-1
    • DSM [Geva & Sitte's Decision Surface Mapping]
  
  
• Clustering, just a few clustering algorithms:
  • Adaptive Clustering Algorithm
  • Batchelor and Wilkins' Algorithm
  • K Means
  • GRASP
  • Sequential Clustering Algorithm
  • ISODATA
  • Similarity Matrix Clustering Algorithm
  
  
• Help
  • On-line help gets the HTML files you're reading just now
  • About... NC, of course.