Impute Missing Values

norita

I am handling Missing Values.

Does this sound reasonable to you? Any thoughts?

I set the parameters in the operator Impute Missing Values as follows:

order: information gain

sort: ascending

Because I want to have the most information based imputation from the dataset. So the ones with least information gain I do first and the ones with a lot of missing at the end.

So I am most convinced to have approached the best when the uncertainty is the most.

Best

Nora

rfuentealba

Hello Nora,

Sorry noone has chimed in.

For what I understand from your questions, you want to generate values. While I think this is a great idea, I would first take some time to see if the values generated (via impute or some other method) is consistent with the other values. If you "can" generate those imputed values successfully, that means these values have a high correlation with others. So it will depend on your data and how much you know about it.

All the best,

Rod.