Replies to This Discussion

Permalink Reply by Ashok B on January 29, 2011 at 10:24pm

Hi,
We are making use of SPSS in our organization, i refer you people to go with SPSS.It is having so many advanced features when compared to other data-mining tools.The main thing about it can handle millions of data together and able to findout churn behaviour of teleco customers.

Regards,
Ashok B.

Permalink Reply by Jozo Kovac on February 1, 2011 at 6:08pm

More important than tool are your data. 

You will need all many attributes about customers + target variable (will / won't churn in next X days).

That's the biggest challenge in front of you. 

 

10M customers? You can always use sampling and reduce it significantly. At least until you find set of the best predictors.

 

All of those tools can handle such job and connection to Oracle isn't a problem. But they are all so different ...

 

Non-free:

KXEN & SPSS Modeler can be mastered very fast.

SAS is great, handles large data very well, may be a little bit more complicated.

 

Free:

I would choose R - great features, but hardest to master - you have to write code, don't have nice GUI.

Weka isn't a real option. RapidMiner is based on Weka, may be an alternative with GUI.

I heard good things about Orange, but haven't ever try. Do it and share your experience.