Logistic Regression Error

Hi,

I have a very simple flow: Read csv -> Change target values to strings -> Take care of missing values (all options accounted for) -> Logistic Regression Learner

I've tried different input files, each between 45k and 150k rows.

Keep getting the same error from the logit learner:

Execute failed: ("NullPointerException") : null

Any suggestions as to what I might be doing wrong?

 

Thank you.

Any chance we can get an example workflow that demonstrates the problem?  Also, there may be some more useful information in the knime log (View > Open KNIME Log).  When you see the error in KNIME you should find a related stack trace down at the bottom of the log file. 

 

Hi Aaron,

I'm new with Knime, as in "this is the very first time I try it", so I'm not sure what you mean by "an example workflow". I have attached the zipped project file to this message, if that is what you mean.

I took a look at the log file, but cannot make much of it. Everything seems OK, except for the error message of course. I've also attached a message with the last few lines from it.

Thank you.

Thanks for the workflow.  For future reference, you can right-click > export, but that worked just fine save for the data file.  May we see that as well?

Aaron

Hi Aaron,

Attached is the data file.

Thank you.

Thanks for the data file.  The problem appears to be related to the number of unique values in a few of the columns in the data set (several over 700).  I have opened a ticket to increase the robustness of the node or at the least give a better error message, but I would also recomend looking into a different classifier such as a decision tree, or tree ensemble, both of which will work better with this sort of data. 

Thank you Aaron, I'll try that.