c4_5testfun.m

常用决策树算法C4.5的实现代码。利用matlab实现。

%C4_5TestFun.m
%Shiliang Sun ([e-mail]), Apr. 8, 2007
%Using the learned 4.5 decision tree to classify samples
%This code is based on the C4_5.m file from "Classification Toolbox for Matlab"
%(http://www.yom-tov.info/cgi-bin/list_uploaded_files.pl). Several bugs have been fixed.

function targets = C4_5TestFun(patterns, tree, discrete_dim)
%Classify recursively using a tree
%  patterns     - Test patterns, (the number of features) * (the number of samples)

indices = 1:size(patterns,2);
targets = zeros(1, size(patterns,2));

if (tree.dim == 0)
    %Reached the end of the tree
    targets(indices) = tree.child;
    return
end

%This is not the last level of the tree, so:
%First, find the dimension we are to work on
dim = tree.dim;
dims= 1:size(patterns,1);
dims(dim)=[]; %ssl

%And classify according to it
if (discrete_dim(dim) == 0)
    %Continuous pattern
    in				= indices(find(patterns(dim, indices) <= tree.split_loc));
    if isempty(dims)
        targets(in)=tree.child(1);
    else
        targets(in)	= C4_5TestFun(patterns(dims, in), tree.child(1), discrete_dim(dims));
    end
    in				= indices(find(patterns(dim, indices) >  tree.split_loc));
    if isempty(dims)
        targets(in)=tree.child(2);
    else
        targets(in)	= C4_5TestFun(patterns(dims,in), tree.child(2), discrete_dim(dims));
    end
else
    %Discrete pattern
    Uf				= unique(patterns(dim,:));
    for i = 1:length(Uf),
        in   	   = indices(find(patterns(dim, indices) == Uf(i)));
        if isempty(dims)
            targets(in)=tree.child(i);
        else
            targets(in)	= C4_5TestFun(patterns(dims, in), tree.child(i), discrete_dim(dims));
        end
    end
end

%END use_tree