[xinunus]

I am looking for some good code examples on how to implement a multi-way decision tree. I have seen examples of binary trees but I have to implement a multi-way decision tree and there doesn't seem to be alot of good examples showing how the algorithm is implemented. These are also called classification decision trees. They are widely used when performing data mining. Any help is appreciated.

Thanks..

This post has been edited by xinunus: 04 July 2008 - 10:21 PM

[lordms12]

You can request this on Request a C++ Tutorial thread.

[Predictor]

xinunus, on 4 Jul, 2008 - 09:15 PM, said:

I am looking for some good code examples on how to implement a multi-way decision tree. I have seen examples of binary trees but I have to implement a multi-way decision tree and there doesn't seem to be alot of good examples showing how the algorithm is implemented. These are also called classification decision trees. They are widely used when performing data mining.

Some existing algorithms do this (ID3 being the classic example), although they tend to split every value out, which is obviously a potential problem for variables with many distinct values (such as numeric variables) or with values that have very low representation in the data.

My question is: What advantage do you perceive in using multi-way splits? Any multi-way tree can be represented as a 2-way split tree, but 2-way splitting algorithms have at once less logistical problems (as mentioned above) and more flexibility.


-Will Dwinnell
Data Mining in MATLAB