PRTools Contents

PRTools User Guide



Classifier evaluation (feature size/learning curve), bootstrap possible


 A Training dataset
 CLASSF Classifier to evaluate
 FEATSIZE Vector of feature sizes  (default: 1:K, where K is the number of features in A)
 TRAINSIZES Vector of class sizes, used to generate subsets of A (default [2,3,5,7,10,15,20,30,50,70,100])
 NREPS Number of repetitions (default 1)
 T Tuning set, or 'bootstrap' (default [], i.e. use remaining  objects in A)

 E Error structure (see PLOTE)


Generates at random, for all feature sizes defined in FEATSIZES or all  class sizes defined in TRAINSIZES, training sets out of the dataset A and  uses these for training the untrained classifier CLASSF. CLASSF may also  be a cell array of untrained classifiers; in this case the routine will be  run for all of them. The resulting trained classifiers are tested on all  objects in A. This procedure is then repeated N times.

Training set generation is done "with replacement" and such that for each  run the larger training sets include the smaller ones and that for all  classifiers the same training sets are used.

If CLASSF is fully deterministic, this function uses the RAND random  generator and thereby reproduces if its seed is reset (see RAND).  If CLASSF uses RANDN, its seed may have to be set as well.



See also

mappings, datasets, clevalb, testc, plote,

PRTools Contents

PRTools User Guide

This file has been automatically generated. If badly readable, use the help-command in Matlab.