data mining - classification tree

[wrijneveld]

Hello,

Is there a way one could build a classification tree in Excel, such as XLMiner could do (but is very costly software).

See e.g. this excel file: http://mande.co.uk/blog/wp-content/uploads/2012/04/epm-data-only.xlsx

and this type of classification tree: http://mande.co.uk/blog/wp-content/uploads/2012/04/epm-tree-2nd1.png

(and a further description of the challenge here: http://www.mandenews.blogspot.co.uk/2012/04/data-mining-tools-and-evaluation.html)


It's not just to reproduce this picture, but to include an optimalisation process in it that finds the smartest way to present this.

regards, wouter

 

[keymaster]

Interesting. I remember developing a classification rule generator in Java way back in 2003 as part of my undergrad project. The tree was made after analyzing lots of cases & would define rules to predict whether a person is diabetic or not.


If your objective is to merely generate the rules (in XML format), you can use VBA. Drawing the tree can be tricky, but do-able. For tree generation, you need some algorithm like C4.5 or ID3. For more on the algorithm, see here: http://www.rulequest.com/Personal/ and http://en.wikipedia.org/wiki/C4.5_algorithm