Hopefully I´m at the right place for my question:

I´m looking for suggestions for models to use to classify multivariate time series. I´m trying to find a way of classifying the behaviour of motors into "good" or "bad" based on current measurments.

I found many possible examples (as found for example in the library sktime) to use, but my biggest problem is that the dataset I have captured is incredibly small because of difficulties in the testing environment. The dataset consist of:

  • 11 full time series of ~2min length
  • a maximum of 12 measured variables for each time series; mainly motor currents

I therefore understand, that a valid classification is probably impossible. Maybe there are models with uncertainty though, that I could use and improve, once I generate more data in the future. Hopefully the uncertainty decreases with more data. That would be a good starting point and the best I can make of the sparse dataset, I believe.

As I understand, Bayesian Classifiers would be similar to what I´m looking for, but they aren´t specific for time series, right?

Any buzzword to look for or model suggestion?



Your Answer

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

Browse other questions tagged or ask your own question.