I'm a biological scientist and I'm trying to interpret a FFT of some biological data.

Background: I have monitored the length of time an animal stands over six days, so every time the animal stands up we have a time code for when the behaviour started and a duration of how long the standing activity lasted for. Looking at this data broken down by day shows us some considerable variation and in the past people have talked about animals not having a 24 cycle but possibly a 26 hour cycle or even a 3 day cycle. I've been told that a FFT can tell me the frequency of the cycle in my data, so I'm trying to discern a pattern in what is essentially six days worth of data.

I've never used FFTs and while I've managed to transform my data in excel, I don't think I'm using the time series correctly which is causing problems when trying to interpret my data.

I have an animal which is standing up for recorded periods of time. For example:

Time ... Duration of activity (seconds)

37 minutes ... 27

174 mins ... 1540

304 mins ... 21

359 mins ... 102

This goes on for 64 values, with the last time record being 7968 minutes. I'm using an excel tutorial I downloaded to try and analyse the data and it's giving me a frequency domain graph with one very high peak at 'zero (Frequency (cycles/time unit)) and a long tail of modulus all the way to the 1/2 frequency.

This is quite different from the example data I've seen in this graph, which tend to have shorter tails.

Now, so far I've just been using a frequency (cycles/time unit) which exists in the prior example which goes 0, 1/64, 1/32...1/2, but as my recordings are not at regular intervals, I'm wondering if this is still correct?

In essence, my questions are:

Am I using the FFT correctly in this instance? I have some questions over whether a

DFT would be better and if so, how should I go about this?

Is the time series 0, 1/64... still applicable in this instance?

How do I interpret the data?

Many thanks!