MATLAB: Decision Tree Classification with 0 on BOTH terminal nodes of branches

decision treefitctree

Using t=fitctree(x,y,'MinLeaf',20) and then view(t,'Mode','Graph') to look at how the tree is working, and I see that in several cases the result at a terminal branch is for BOTH end points to be '0', or BOTH to be '1'.

How can this be? I would expect each branch to end with one '0' and one '1', never two '0' or two '1'.

Thanks!

Best Answer

Nothing wrong with that. A classification tree optimizes the Gini index and can split a parent node in two child nodes dominated by the same class.

Related Solutions

MATLAB: Decision trees, only binary branches

If you fit a classification tree to the famous Fisher iris data, you get this:

>> load fisheriris
>> f = fitctree(meas,species);
>> view(f)
Decision tree for classification
1  if x3<2.45 then node 2 elseif x3>=2.45 then node 3 else setosa
2  class = setosa
3  if x4<1.75 then node 4 elseif x4>=1.75 then node 5 else versicolor
4  if x3<4.95 then node 6 elseif x3>=4.95 then node 7 else versicolor
5  class = virginica
6  if x4<1.65 then node 8 elseif x4>=1.65 then node 9 else versicolor
7  class = virginica
8  class = versicolor
9  class = virginica

So you can see:

The tree has more than two branches
It can predict more than two classes (three in this case)
However, at each branch it makes a binary decision, X<s or X>=s

If X happens to be a categorical predictor, then it still makes a binary decision of the form "X is among this group of categories" or "X is among this other group."

MATLAB: How to delete the branches

Maybe I can delete these segment lines in ArcGIS as the small segment lines can be identified as short length.

Best Answer

Related Solutions

MATLAB: Decision trees, only binary branches

MATLAB: How to delete the branches

Related Question