cpann_class_param.m

this file is leverage algorithm written in matlab as m-file and tested in matlab.so anyone can ue th

function class_param = cpann_class_param(class_calc,class)

% cpann_class_param calculates classification parameters
% (error rate, non-error rate, specificity, precision and sensitivity)
%
% class_param = cpann_class_param(class_calc,class);
%
% input:
%   class           class vector [1 x n]
%   class_calc      calculated class vector [1 x n]
% 
% output:
%   class_param     structure containing confusion matrix, error rate, non-error rate, 
%                   specificity, precision and sensitivity
%
% see the HTML HELP files (help.htm) for details 

num_class = max(class);
nobj = size(class,1);

conf_mat = zeros(num_class,num_class+1);
for g = 1:num_class
    in_class = find(class==g);
    for k = 1:num_class
        conf_mat(g,k) = length(find(class_calc(in_class) == k));
    end
    conf_mat(g,num_class + 1) = length(find(class_calc(in_class) == 0));
end

% error rates
ner     = sum(diag(conf_mat(:,1:end-1)))/nobj;
not_ass = sum(conf_mat(:,end))/nobj;
er      = 1 - ner - not_ass;

% sensitivity, specificity, precision
for g = 1:num_class
    if sum(conf_mat(:,g)) > 0
        precision(g)   = conf_mat(g,g)/sum(conf_mat(:,g));
        sensitivity(g) = conf_mat(g,g)/sum(conf_mat(g,:));
    else
        precision(g)   = 0;
        sensitivity(g) = 0;
    end
    in = ones(num_class,1); in(g) = 0;
    red_mat = conf_mat(find(in),:);
    specificity(g) = 0;
    for k = 1:size(red_mat,2)
        if k ~= g; specificity(g) = specificity(g) + sum(red_mat(:,k)); end;
    end
    if sum(sum(red_mat)) > 0
        specificity(g) = specificity(g)/sum(sum(red_mat));
    else
        specificity(g) = 0;
    end
end

class_param.conf_mat = conf_mat;
class_param.ner = ner;
class_param.er  = er;
class_param.not_ass = not_ass;
class_param.precision = precision;
class_param.sensitivity = sensitivity;
class_param.specificity = specificity;