I am at the end of my study (almost 6 months). I need time to edit my English (my supervisors do not want to help with that. It is Ok for me.). However, they asked me to add extra computation to my study, which is not directly relative to my topic. I tried to tell them that but they still want me to do so. I do what they asked me to do just to show them that I respect their suggestions and comments. However, for more than one time, their idea did not fit my data. This is not my problem as, for their comment, I need to use an existing package. The result is very acceptable as the existing model does not fit such type of data (my data). The model they asked me to apply to my data is already known to be inappropriate to deal with a specific type of data. Therefore, several works have been developed to cover this problem. If I keep going with this inappropriate model, then I will lose my time. Therefore, I would like to send my supervisors an email showing them that the model is not appropriate for my data and result in an inaccurate result and report errors. However, I do not like to make them feel that their idea is not good. So, how could I send a polite message to my supervisor?
My experience is that emails are far less effective than face-to-face interactions. If you are at the point that an email is all you can do because the relationship is dysfunctional and you just want to get out of there, then a simple statement acknowledging the validity of the approach in general but not in this domain, with the reminder that you need to graduate in a bit, would do the trick. Your advisors have no interest in needlessly prolonging your PhD (at least unless they have a serious grudge against you).
Thank you for suggesting method Y. I think it's a potentially useful approach that I am happy to explore in detail; however, it would seem that prior work indicates that the approach may not be as effective as we believe in domain Z (see attached PDF of the study by Alice and Bob Smith on the topic). I will be happy to add a discussion on this to the thesis (in the relevant section), but I do not believe that running a detailed analysis will yield fruitful results given the current state of the art.
Given that I have to submit my thesis in 6 months, I think that the best course of action would be to focus on the results we have now, rather than expanding in new, potentially unproductive, directions.