Testing KNIME with KDD Cup 2009 data

KNIME Analytics Platform
1

Hello,

This is hte very first time I'm using KNIME and I thought KDD cup 2009 could be a good benchmark.

As a first trial, I'm just trying to create a logistic regression model.

I was expecting to find in the Logistic Regression Learner an option to apply a basic feature selection procedure like stepwise but it doesn't have.

I believe this is done by the Meta node "Feature Elimination". When I execute this node, after replacing the Bayes Learner inside by the logistic regression, I get the following error:

ERROR     Logistic Regression Learner        Execute failed: ("NullPointerException"): null

You may find attached the workflow. The data is 28 Mb so I can't attach.

It can be found here:

http://www.sigkdd.org/kdd-cup-2009-customer-relationship-prediction

Thanks a lot for any assistance!

Kind regards,

Joel

 

 

2

Hi Joel, 

Nice to meet you.  I only see the data, no workflow.  Did you forget the attachement?

 

3

Hi Aaron, Nice to meet you too. Thanks for replying. Well, I was convinced having done it but I probably didn't pick the right stuff. This time I zipped all that seems as representing the workflow to me. Let me know if it's right this time. thanks again, Joel

4

Hi Aaron,

I already provided the workflow. If there is any chance you can find some time to take it look, I would highly appreciate. It's a small workflow and the problem must be with my novice level and not with Knime...

thanks a lot for this initial support.

Kind regards,

Joel