Fill in Missing Values by average of whole row
I'm trying to fill in the missing values by averaging the rows or possibly use some row level computation like regression. I'm wondering if i need to do un-pivot it? I would greatly appreciate a sample process that can do this . Below is my sample row level data
Find more posts tagged with
Sort by:
1 - 5 of
51

SGolbert
New Altair Community Member
Updated by SGolbert
Hi @msacs09
You can do it with the Generate Aggregation operator and a Loop. Something like this:
It is a bit clumsy, but I didn't find an easier way.
Regards,
Sebastian
@SGolbert Thank you. If I may ask, how do i replace the average here with a regression model such as KNN or some other learner, so that we impute with a regression value as opposed to average, such that we do not repeat the SAME value for every missing row for that ID?
Thank you for your valuable time.
Thank you for your valuable time.
Hi msacs,
you could do something very similar, but just use a prediction column (after applying the model) instead of the row_mean column. Another option is the Impute Missing Values operator, that does practically the same, all in one go.
Do you want to try it out?
Regards,
Sebastian
It appears that the imputation operator has a bug. It complains about missing exampleset as reported here. So I was curious if your approach would resolve that issue.
https://community.rapidminer.com/discussion/54470/imputing-challenge-with-a-best-possible-ensemble#latest
https://community.rapidminer.com/discussion/54470/imputing-challenge-with-a-best-possible-ensemble#latest