I have some results that I unfortunately had to gaussian filter due to noise. This is to simplify post-processing later on, like first derivative.

I have asked around and I've been told that if it is honestly mentioned in the paper, this is not a problem. I still have some questions:

  • Because of the filtering applied I would like to do a conclusion on this data without looking at the numerical values but at the trends because they are mostly non-affected by the filtering. When plotting should I then omit the x and y number 'ticks' to stress that numerical values do not make sense? (I feel like putting them would be cheating, because they are somewhat still good in my case)

  • Should I still show the original data in the paper?

1 Answer 1


When you publish data that you use to make an argument, it is crucial to describe how the data was obtained. This would usually also include any preprocessing or smoothing, even if it is only applied for data in plots.

Your suggestion regarding also showing unaltered data is a good idea: For regression or box plots scattering the original data over the model can be insightful.

Also, this can be very problem-dependent so I would advise to check out how researchers in your field usually handle the same kind of data and problem.

  • 2
    Somewhat field dependent, but in many subjects it's best practice to provide the raw data (barring ethical/legal obstructions) along with a clearly written computer script that takes the raw data as input and produces the filtered data as output.
    – user176372
    Dec 31, 2023 at 0:11

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .