I have a csv data file which contains a non-numeric column (tarif) which can have hce, hp, ....

The problem is some missing values. As the values are non-numeric (hce or hp), it is impossible to impute with mean().

Can you suggest ideas to resolve this problem to fill empty columns?

  • $\begingroup$ How to approach this in any specific software is off-topic here: see stats.stackexchange.com/help/on-topic for guidance on software-related topics. Otherwise there is a statistical question here. A possibility is to use some kind of logit or multiple logit model for imputation. It's not clear from your post whether there are only two possible values or several. (Empty columns are insurmountable; I guess you mean empty cells, or whatever Python calls them.) $\endgroup$ – Nick Cox Sep 5 '16 at 16:38
  • $\begingroup$ @NickCox Thank you for you quick reply. I have three possible values : either "HC", "HP" or "PTE". Yes I mean empty cell :) $\endgroup$ – Poisson Sep 5 '16 at 16:44
  • $\begingroup$ Imputation with multiple logit then. If that's too complicated, then ignoring rows (or whatever Python calls them) with missings is indicated. $\endgroup$ – Nick Cox Sep 5 '16 at 16:48
  • $\begingroup$ Ok, Have a you any tutorial ti suggest to me concerning Imputation with multiple logit ? thanks $\endgroup$ – Poisson Sep 5 '16 at 16:54
  • $\begingroup$ I'd Google or find a textbook on multiple imputation. I don't ever do it; I just know about it. If you've never heard of it, it won't be an easy job. $\endgroup$ – Nick Cox Sep 5 '16 at 17:20

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.