SDTREE Decision tree classifier

    P=SDTREE(DATA)

    P=SDTREE(P,ACTIVE)  % set the number of active thresholds

 INPUT
   DATA    Data set

 OUTPUT
   P       Decision tree pipeline

 OPTIONS
  'full'     Grow full tree without pruning
  'levels',N Limit the number of tree levels (default: grow full tree
             and prune it using error on the validation subset)
  'no pruning'   Do not prune the tree
  'test',TSDATA  Provide separate set for pruning the tree
  'trfrac'   Fraction of data used for training/validation split (def: 0.8)
  'minsamples' Minimum number of samples in a node to stop further
               splitting (default: 10)
  'maxnodes'   Maximum number of nodes (default: 1000)
  'thr'      How to generate thresholds (default: 'grid' or 'samples')

 DESCRIPTION
 SDTREE implements decision tree training. By default SDTREE splits the
 available DATA into training and validation parts. Then it builds the
 tree on the training part and estimates error using the validation
 subset.  Tree with minimal validation set error is returned.  SDTREE is
 scalable to very large numbers of samples.

 To grow full tree without pruning, use:
 >> p=sdtree(data,'full')
 'full' is synonym for: sdtree(data,'no pruning','minsamples',1,'thr','samples')

 READ MORE
 http://perclass.com/doc/guide/classifiers.html#sdtree

sdtree is referenced in examples:

• kb22: Note on decision tree performance and speed