Logistic Regression Error Message

[RNIB_ianhill]

Hi there,

I’ve been attempting to run a logistic regression model, Initially I was getting an error message pertaining to the algorithm not converging. I’ve resampled the data so it a 50/50 split responder / non responder. I’m now attempting to run the model using the Iteratively reweighted least squares algorithm. I’ve tried running the model with just two variables. I’m now getting the following error message please see the file attachment .

Thank you.

Ian

[thor_landstrom]

Hey @RNIB_ianhill,

It is hard to single out what may be causing the issue, are you able to attach your wf? It is probably related to your input data and it may need additional preprocessing.

I can see you have a DB Query, so if you could have like a sample set saved as a knime table that would be very helpful.

TL

[Brock_Tibert]

Echoing @thor_landstrom, it’s not really possible to debug your solution with a screenshot. To extend what might be an issue, check for:

• Missing data in the feature or target
• constant values in a column included in your dataset fed to the model

[RNIB_ianhill]

Hi Thor / Brock,

Thank you for the response below. I think I have identified the issue. I’ve create new variables in Microsoft SQL Server. I’ve used the column filter node to exclude these variable and no longer encounter this error message.

Regards.

Ian