Generates a confusion matrix according to true and predicted data labels. CM(i,j) denotes the number of elements of class i that were given label j. In other words, each row i contains the predictions for elements whos actual class was i. If IDXpred is perfect, then CM is a diagonal matrix with CM(i,i) equal to the number of instances of class i. To normalize CM to [0,1], divide each row by sum of that row: CMnorm = CM ./ repmat( sum(CM,2), [1 size(CM,2)] ); USAGE CM = confMatrix( IDXtrue, IDXpred, ntypes ) INPUTS IDXtrue - [nx1] array of true labels [int values in 1-ntypes] IDXpred - [nx1] array of predicted labels [int values in 1-ntypes] ntypes - maximum number of types (should be > max(IDX)) OUTPUTS CM - ntypes x ntypes confusion array with integer values EXAMPLE IDXtrue = [ones(1,25) ones(1,25)*2]; IDXpred = [ones(1,10) randint2(1,30,[1 2]) ones(1,10)*2]; CM = confMatrix( IDXtrue, IDXpred, 2 ) confMatrixShow( CM, {'class-A','class-B'}, {'FontSize',20} ) See also CONFMATRIXSHOW Piotr's Computer Vision Matlab Toolbox Version 2.12 Copyright 2014 Piotr Dollar. [pdollar-at-gmail.com] Licensed under the Simplified BSD License [see external/bsd.txt]

Generated by