classification/prediction model with one single label attribut vs. multiple label attribut

LeMarc

Hi,

I build two different classification models to compare.(Example sets to build and apply the model in both cases are the same):

1. With a single label

2. With mutliple labels (using the multi label modeling operator)

First Scenario: 1 Label attribut is selected for both models to be predicted/classify. The output of both example set differs. Results of the second model has some empty fields for all attributes [except the prediction role] of the data set.  why is it?

1. Model

 

2. Model with missing values

Second Scenario: If I select two attributes to be predicted in the second model. The number of "wrong predictions" of Business Units increase from (7 to 10).  Also there is an empty value in row 10 which does not exist in the original data set. What are the reasons behind that?

Thank you!

Telcontar120

We cannot see your data, but this error means that you only have one label value, like "Yes" for every record. Logistic Regression works on binnominal classification problems so you need to have a label with exactly two values, like "Yes" and "No."

Telcontar120

We cannot see your data, but this error means that you only have one label value, like "Yes" for every record. Logistic Regression works on binnominal classification problems so you need to have a label with exactly two values, like "Yes" and "No."

LeMarc

Thanks for the answer though a little bit late!