1
$\begingroup$

What is meant by frequency of a time series? I am having data observations of energy consumption from 01/05/2016 - 31/05/2017 such that there are 96 values per day. What will be the frequency of the time series in such case?

$\endgroup$
  • 1
    $\begingroup$ Do you refer to the frequency of the signal or the sampler? $\endgroup$ – Emre Jul 11 '17 at 19:52
  • $\begingroup$ Basically means how much time elapses between two data points. If someone says We have 5-minute frequency data, that means they got a data point every 5 minutes. In your example 96 values per day (assuming 24 hours of operation of the sensor) would mean (24 hours * 60 mins = 1440 minutes per day. You have 96 points per day, meaning a point was measured every 1440 / 96 = 15 minutes. 15-minute data! $\endgroup$ – n1k31t4 May 7 '18 at 17:00
1
$\begingroup$

The sampling frequency, or sample rate, is the number of equal-spaced samples per unit of time. For instance, if you have 96 equally spaced observation per day, then you sampling rate is 96/day, or 96/24/3600=0.0011 Hz. Hz, which means per second, is widely used for sample rate.

The frequency of your data is different. I you signal is periodic or approximately periodic (say, the similar energy consumption patterns repeat everyday), the the frequency is the number of occurrence of that repeating pattern per unit of time. More generally, if your data does not look like periodic, it would help to conduct Fourier Analysis to understand the frequency components in your data. The Fourier Analysis (Fourier Transform) is the operation to transform time-domain data into the frequency domain.

$\endgroup$
0
$\begingroup$

I don't think that this is defined in general. So you should be more specific on where you are using this.

I could guess, maybe you mean the Time-Series Object in R? Then take a look at: https://stat.ethz.ch/R-manual/R-devel/library/stats/html/ts.html

And here the definition is:

frequency -> the number of observations per unit of time.

$\endgroup$
0
$\begingroup$

It's mainly defined by how you perceive your data and the way it's sampled, but basically you can define a time-series frequency as: (the number of samples) per your (sample unit)

In your case: 96 per day, since it seems that a "day" is your meaningful sample-unit.

If you had several months, or even better - years, you could argue your frequency to be 96*30 (monthly # of samples) and the sample unit as months

$\endgroup$

Your Answer

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

Not the answer you're looking for? Browse other questions tagged or ask your own question.