|
Starts the graphical user interface of the hierarchical cluster analysis (HCA) and instantiates the class TFrmHCA. The user interface can be controlled by using the global instance FrmHCA.
The parameter LinkageType specifies the type of linkage used for calculating the dendrogram:
1 ... Single Linkage
2 ... Complete Linkage
3 ... Average Linkage
4 ... Ward's Method
Likewise the parameter DistMeasure determines the distance measure used for the HCA:
1 ... Euclidean
2 ... Squared Euclidean
3 ... Manhattan
The parameter ReducedData allows to speed up calculations for preliminary analyses vy using a reduced dataset. The reduction of the dataset works according to the following idea: in the first step the distances of all spectra (pixels) are calculated and those which are closer to each other than a predefined part are combined into mini clusters which are represented by their average spectra. This way the the number of spectra submitted to the final cluster analysis is reduced to a manageable number (less than 10000). The sensitivity of the creation of the mini clusters can be controlled by the parameter PercMinDist which defines the percentile of all distances which are not regarded significant. A higher value of PercMinDist creates less mini clusters.
The function returns the following error codes:
0 ... everything is ok
-1 ... invalid linkage type
-2 ... invalid distance measure
|
Script Language 
