Example set has no nominal label: using shuffled partition instead of stratified partition!

User111113
User111113 New Altair Community Member
edited November 5 in Community Q&A
I am getting this error while running my model as a warning actually and I tried to change data type of one field from real to nominal but this warning still didn't go away. I am using GBT for prediction for my data set.

Can you tell me how to resolve this issue. thank you
Tagged:

Best Answer

  • BalazsBarany
    BalazsBarany New Altair Community Member
    Answer ✓
    Hi @User111113,

    this warning applies to the attribute marked as label. This is the one you're predicting.

    Stratified sampling is only available for nominal labels, for classification problems. If your label is numeric, then you're doing regression, which is fine.

    Stratified sampling (e. g. in Split or Cross Validation) means keeping a similar ratio of the target classes in the test and the training sets. This is not available with regression, but that's OK.

    If you set the operator that is now set to stratified sampling to shuffled, the warning should disappear.

    Regards,
    Balázs

Answers

  • BalazsBarany
    BalazsBarany New Altair Community Member
    Answer ✓
    Hi @User111113,

    this warning applies to the attribute marked as label. This is the one you're predicting.

    Stratified sampling is only available for nominal labels, for classification problems. If your label is numeric, then you're doing regression, which is fine.

    Stratified sampling (e. g. in Split or Cross Validation) means keeping a similar ratio of the target classes in the test and the training sets. This is not available with regression, but that's OK.

    If you set the operator that is now set to stratified sampling to shuffled, the warning should disappear.

    Regards,
    Balázs

Categories