classifierEval

Evaluation of a given classifier

Contents

Syntax

Description

computedClass=classifierEval(classifier, DS, cPrm) returns the computed class of the dataset DS on a given classifier.

• classifier: a string specifying a classifier
  • classifier='qc' for quadratic classifier
  • classifier='nbc' for naive Bayes classifier
  • classifier='gmmc' for GMM classifier
  • classifier='knnc' for k-nearest-neighbor classifier
  • classifier='linc' for linear classifier
  • classifier='src' for sparse-representation classifier
  • classifier='svmc' for support vector machine classifier
  • classifier='lsec' for least-square estimate classifier
• DS: data set for training
• cPrm: parameters for the classifier, where cPrm.class(i) is the parameters for class i.
• computedClass: a vector of computed classes for data instances in DS

computedClass=classifierEval(classifier, DS, cPrm, plotOpt) also plots the decision boundary if the dimension is 2.

[computedClass, logLike, recogRate]=classifierEval(...) returnsmore info, including the log likelihood of each data instance, and the recognition rate (if the DS has the info of desired classes).

Example

DS=prData('3classes');
plotOpt=1;
classifier='qc';
[cPrm, logLike, recogRate, hitIndex]=classifierTrain(classifier, DS);
DS.hitIndex=hitIndex;		% Attach hitIndex to DS for plotting
classifierPlot(classifier, DS, cPrm, 'decBoundary');

Published with MATLAB® R2018a