How to create readable decision trees

I am trying to produce a binary decision tree as following:

% Dataset
r2 = [0;0;1;0;1;0;0;1];
r3 = [0;0;0;0;0;1;1;0];
r4 = [0;1;1;1;1;1;1;0];
Group = [1;1;1;1;1;2;2;3];
T = table(r2,r3,r4,Group)

T = 8×4 table
    r2    r3    r4    Group
    __    __    __    _____

    0     0     0       1  
    0     0     1       1  
    1     0     1       1  
    0     0     1       1  
    1     0     1       1  
    0     1     1       2  
    0     1     1       2  
    1     0     0       3  

% 'Group' is the target in dataset T
model = fitrtree(T, 'Group', 'MinParentSize', 1);

% View the model as a graph
view(model,'Mode','graph');

The decision tree that MATLAB gave me is shown in MatlabTree.png. I also produced a tree for the same dataset using Latex and that is shown in LatexTree.png.

I want MATLAB to display parent nodes same as Latex tree's parent node. For example, Matlab's tree says "r3>0.5 r3>=0.5", which is not as good to read as Latex generated tree. Can this Matlab tree be made more readable like the Latex one?

Answer 1

[Christopher McCausland]

Hi Awais,

I had a similar problem with Markov chains recently. I ended up using digrapgh() to solve the problem, that being said its still a lot of work.

Kind regards,

Christopher

[Awais Saeed]

Thanks for the suggestion @Christopher McCausland. digrapgh() is not as easy, efficient and exact as fitrtree(). I searched on the internet and it seems that the appearance cannot be changed. Let's hope that MATLAB will offer this ability in future releases.

[Christopher McCausland]

Hi Awis,

Digraph isn't as easy you are correct, it does look good when its finsihed though. To make it quicker you can call parameters already created in 'model' Esentially you create model, use graphplot() to generate your nodes and lines and then use digraph() and plot() to generate the new plot based on the inputs from 'model'.

I hope you can get something sorted, I will have a go myself later and see if I can get anything!

Christopher