0
$\begingroup$

I have a dataset about a bunch of users and their comment on Instagram, each row is a comment. Some of the columns of this table are related to the users and some of them have multiple comments.

I'm doing the EDA and plotting my charts to see some viz about the data but when looking at follower_count for example, do I need to subset the data and drop the duplicates before plotting?

I guess it depends and both are valid, but I'd be happy to hear other's opinion on this.

$\endgroup$

1 Answer 1

0
$\begingroup$

If you drop the duplicates then would that not alter the distribution of your input data? That in turn would lead to fitting a posterior distribution to a different dataset than the original one. Dropping duplicates should likely be done only if they were input data errors - for example the data ingestion included one of the files twice.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.