what about if your data acquisition rate was variable (dt not constant)?
Then you really can't make a good filter based on signal frequency. You could approximate (ie average) the sample rate and use that to come close. Or perhaps a simple moving average would be good enough.
[Edit: started a reply, got interrupted, and found that crossrulz had already answered similarly. Original post below, for the record.]
Can you give some more details (and maybe start a new thread, as this one's been marked as solved, which will cause many readers to skip over it)?
The filter functions will calculate coefficients that depend on and assume constant sample rate data. If you apply those coefficients to variable sample rate data, you won't get the standard filter frequency response characteristics. You'll get something else.
In the most common case of lowpass filtering, the "something else" you get will probably still behave somewhat like a lowpass filter, but with unpredictable frequency domain characteristics.
Do you know what all the dt intervals are? If so, you can do some interpolation to construct *apparent* sample values at a constant sample rate. You could filter *that*, thereby getting the know filter behavior. If desired, you could do another interpolation to get back to the original variable dt intervals.
The previous reply of crossrulz is correct. Also, I was thinking a little more on my problem, and I realised that I was doing everything wrong. What I have is a set of recorded data, at different sampling rates, for example, the first initial 60 points at 1 point every 60 seconds, the next 1000 points at 1 point every second, the next 60 points at 1 point every 60 seconds ....... and so on.
In any case, I was playing with the filters of the equipment where I acquire the data using low pass filters at different cutoff frequencies (10, 100, 1000, none). However, my equipment is limited to these frequencies only. So, I was thinking to apply the Labview digital filters in my recorded signal when I use no filter in my equipment.
I think that my original idea is fundamentally wrong, because, it is incorrect to apply frequency filters in already recorded data.
Anyway, thank you for your interpolation idea, it is always a useful tool.