Time Series pattern recognition and classification problem



I have some labeled sensor data.

Now, I would like to know how to extract features from time series using DFT, DWT, and HAAR transforms.

I know that the transformations above transform a signal to frequency or time frequency domain, but how can I get average frequency for this non stationary signal?



Posted 2017-03-15T07:21:17.103

Reputation: 51

Have you been able to transform your data? Or do you basically ask how to transform it? – Nikolas Rieble – 2017-03-18T12:09:43.170

I tried to edit your post but in the last sentence I got stuck! I changed "the" to "but" but I think it's still unclear what you want. – Kasra Manshaei – 2017-03-24T09:55:00.020

1Not sure why you would want to find an average frequency. What is your higher-level question you want to answer? What kind of sensor is it for which you have data? What is your sampling rate? – Adrian Keister – 2017-06-08T16:37:17.380



I do not fully understand what you mean by "average frequency". But this may help to think about it.

  • You can calculate the power spectrum of the signal using DFT.
  • From the power spectrum, you will be able to get the average power of all the frequencies.
  • Use this value of average power to map the frequency which it closely corresponds to.

This will give a sort of the average frequency you are looking for from the average value of power.

Arun Aniyan

Posted 2017-03-15T07:21:17.103

Reputation: 183