PRTools Contents

PRTools User Guide



Error/performance estimation by cross validation (rotation)


 A Input dataset
 CLASSF The untrained classifier to be tested.
 NFOLD Number of folds  (default: number of samples: leave-one-out)
 NREP Number of repetitions (default: 1)
 TESTFUN Mapping,evaluation function (default classification error)

 ERR Average test error or performance weighted by class priors.
 CERR Unweighted test errors or performances per class
 NLAB_OUT Assigned numeric labels
 STDS Standard deviation over the repetitions.
 R Index array with rotation set


Cross validation estimation of the error (defined by TESTFUN) of the  untrained classifier CLASSF using the dataset A. The set is randomly  permutated and divided in NFOLD (almost) equally sized parts, using a  stratified procedure. The classifier is trained on NFOLD-1 parts and the  remaining part is used for testing. This is rotated over all parts. ERR is the weighted class error avaraged over the class priors. CERR are the  class error frequencies. The inputs A and/or CLASSF may be cell arrays of  datasets and classifiers. In that case ERR is an array with on position  ERR(i,j) the error of classifier j for dataset i. In this mode CERR and  NLAB_OUT are returned in cell arrays.

For NREP > 1 the mean error(s) over the repetitions is returned in ERR and the standard deviations in the observed errors in STDS.

If NREP == 'DPS', crossvalidation is done by density preserving data  splitting (DPS). In this case NFOLD should be a power of 2.

In case CLASSF == [] an [NREPS,M] index array is returned pointing to a  fold for all M objects. No training or testing is done. This is useful  for handling training and testing outside PRCROSSVAL.

Inside PRCROSSVAL warnings generated by PRWARNING are temporarily  suppressed.


1. R. Kohavi: A Study of Cross-Validation and Bootstrap for Accuracy Estimation and Model Selection. IJCAI 1995: 1137-1145.
2. M. Budka, B. Gabrys, Correntropy-based density-preserving data sampling as an alternative to standard cross-validation, IJCNN2010, 1-8

See also

datasets, mappings, dps, cleval, testc,

PRTools Contents

PRTools User Guide

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