The unit DENDROGRAM performs a hierarchical cluster analysis and displays the result as a dendrogram. Both the clustering algorithm and the visual appearance can be adjusted in many aspects.

Among the implemented clustering methods are single linkage, complete linkage, Ward's method, average linkage, and unweighted pair group with arithmetic mean. The user has a choice of several distance measures, such as the Euclidean distance, the city-block distance, the Tanimoto coefficient, or der Dice coefficient.

The application of TDendrogram is straightforward: simply store your data in the internal data container of the component, and the data will be clustered automatically. The results can be panned, zoomed and displayed in various directions with various colors. More details can be found in the online help pages.

The unit DENDROGRAM is part of the SDL Component Suite.

Dendro

This example application generates two clusters of random numbers and displays the results of the cluster analysis as a dendrogram. The user may change various parameters of the system.

dendro.zip (327 kB)

What's new:

• Release 9.0   [May-28, 2007]
   
  • no changes
  
  
• Release 8.5   [Feb-18, 2006]
   
  • TDendrogram now supports class colors
  • new method CalcClasses allows to assign class numbers to the data which are derived from the cluster analysis
  
  
• Release 8.2   [Nov-03, 2004]
   
  • no changes
  
  
• Release 8.1   [Aug-29, 2004]
   
  • available for Delphi™ 8 for .NET
  • methods CopyToBMP, CopyToClipboard, and CopyToBitmap implemented
  
  
• Release 8.0   [Apr-15, 2004]
   
  • first release to the public