I am looking for an alternative way of analysing my data which will be used for ML. I am using Matlab to implement my code. Using my code, I obtain arrays which are used to describe signals. Each array represents a new and different signal.

I have used the corrcoeff method in Matlab and it delivered what I needed. But, I need another alternative way to verify the correlation between all the signals. Is there a better way to do this which is easy to interpret?

Dumb question: is it better to compare the images of the signal generated by Matlab or to compare the arrays?

Forgive me if this sounds badly formulated but I hope it is clear.

Thank you in advanced!

  • $\begingroup$ If you add a snippet of what your arrays look like, you might get more/better answers. Currently, it is unclear what your array looks like (shape?) $\endgroup$ – Romain Reboulleau Oct 7 '19 at 5:12
  • $\begingroup$ I currently don't have access to my array right now. But I can tell you it is a 50 x 12 array. 50 different data with 12 fields. All of this data will be compared to the initial data and will output a correlation coefficient. I am now thinking how i can best graphically display these data to have a good overview of all the coefficients and group them together $\endgroup$ – Mike Oct 13 '19 at 22:17

Based on the current details about your question, I'm not sure what you seek, but I will give it a try.

I assumed you already tried Pearson's coefficient. There are other metrics, which are more or less difficult to implement if not already existing in matlab.

All those will provide some insight about dependencies within your dataset.


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.