Error: Label Not Present

varunm1

Hello,

One of my students is using rapidminer on Mac Pro (Mojave) Java 8. The process has a label in the dataset and still its throwing an error stating "Input example set doesn't have label".

Label column in dataset is label and assigned during data loading into repository. I checked the process and it shows label column

I attached the process .rmp, dataset and log file.

@rfuentealba @sgenzer

Thanks for your help

rfuentealba

Hello, my dear @varunm1

I didn't have the same repository structure with me, but I simply removed that one, added a reader for the provided Excel file, added a "Set Role" before the multiplication of parameters, and the Cross Validations passed without problems.

See attached process and a picture:

Also the log indicates that the data didn't have a label before RapidMiner executed the three stooges cross validations. It actually switched from stratified sample to shuffled sample because it didn't have a way to reorganize the data in function of the labels.

I marked the line that indicates it on your provided log file.

Perhaps the process of changing the .ioobject containing the data wasn't saved?

There are two easy fixes. One is to add the Set Role operator right after reading your data. The other one is opening your .ioobject with the Data Editor panel and make fixes accordingly.

All the best,

Rod.

lionelderkrikor

@varunm1,

I tested by importing your Excel file (via "import data") and by setting Label as "label"  and changing Label type as "Binominal" during import and the process works fine...

To see if, in your case, it is linked to Logistic Regression can you disable the CV associated to Logistic Regression in order
to see if the other CV are executing without bugs ???

Thanks 

Regards,

Lionel

lionelderkrikor

@varunm1,

The process you shared has no defined Label (no Set Role operator in the process)....

By adding a Set Role and set the attribute "Label" as label, the process works fine....

Hope this helps,

Regards,

Lionel

PS : I'm using RM 9.4 / Windows 10

 

varunm1

Sorry @lionelderkrikor. I updated the thread, label is set during data loading and i changed type to binominal as well

rfuentealba

Hello, my dear @varunm1

I didn't have the same repository structure with me, but I simply removed that one, added a reader for the provided Excel file, added a "Set Role" before the multiplication of parameters, and the Cross Validations passed without problems.

See attached process and a picture:

Also the log indicates that the data didn't have a label before RapidMiner executed the three stooges cross validations. It actually switched from stratified sample to shuffled sample because it didn't have a way to reorganize the data in function of the labels.

I marked the line that indicates it on your provided log file.

Perhaps the process of changing the .ioobject containing the data wasn't saved?

There are two easy fixes. One is to add the Set Role operator right after reading your data. The other one is opening your .ioobject with the Data Editor panel and make fixes accordingly.

All the best,

Rod.

lionelderkrikor

@varunm1,

I tested by importing your Excel file (via "import data") and by setting Label as "label"  and changing Label type as "Binominal" during import and the process works fine...

To see if, in your case, it is linked to Logistic Regression can you disable the CV associated to Logistic Regression in order
to see if the other CV are executing without bugs ???

Thanks 

Regards,

Lionel

varunm1

Final update, @rfuentealba informed that the user might have created Label (by typing) instead of selecting label from drop down and yes that is the reason. It took as user defined data type instead of Label. Thanks for quick resolution