Guys,

I am trying to do the standard Customers / Bike Shop data mining
exercise. I have both a Decision Tree and Naive Bayes model and both
work fine. When I process and then go to the Mining Accuracy tab, my
book tells me that I should see 4 lines in the Lift Chart:

1. The "Ideal" line
2. The Decision Tree line
3. The Naive Bayes line
4. The "Random" or "Random Guess" line.

However, I can get the Random line to show up! Everything works, just
this line is missing and it does not appear in the Mining Legend
either.

Has anyone seen this before? Kind of hard to determine my "lift"
without it.

Thanx!

Sincerely,

J'son

I guess you did not specify the state of the predictable column. Take a look
at both versions of Lift Chart in Books OnLine, "Lift Chart" topic at
ms-help://MS.SQLCC.v9/MS.SQLSVR.v9.en/uas9/html/ab77eca1-bd48-4fef-b27f-ff5b648e0501.htm.

--
Dejan Sarka
http://www.solidqualitylearning.com/blogs/

Thanks Dejan... when I select a value in the Predictable column (say
"Mountain Bikes") the Random Guess line does appear in the Lift Chart,
but why is that necessary? I thought that I could see the "lift" of my
models when looking at all discreted values in my Predictable column.

The book I'm currently using - Applied Microsoft Analysis Services 2005
by Teo Lachev (ProLogika) - shows the Random Guess line appearing with
or without a value in the Predictable column.

Thanks for your help and the link!

J'son


Dejan Sarka wrote:

The random Guess line was there in beta / CTP versions even without
predictable value, and the book was written using these versions. I cannot
say why this was changed - maybe somebody from MS will drop in.

--
Dejan Sarka
http://www.solidqualitylearning.com/blogs/