I'm working on a time-series classification problem. There are 3 classes. Dataset consists of 6 sequences from each class (total 18 sequences). Each sequence is 19,000 in length. What are some time-series classification approaches I can look into?

  • $\begingroup$ if the long sequences contain enough information in the length to distinguish the classes then any seuqence classification model may work (eg LSTM) $\endgroup$
    – Nikos M.
    Mar 22, 2021 at 18:38
  • $\begingroup$ @NikosM. What would be an appropriate way to verify if the sequences contain enough information? For example, say I have sensor bootup data (voltage) from three different types of sensors. The sequences look very similar but there are differences. $\endgroup$
    – siaabd001
    Mar 23, 2021 at 5:58
  • $\begingroup$ Only way to try and use them and maybe see the time correlations of each sequence with each other in the same class vs with other classes. This could give you a hint $\endgroup$
    – Nikos M.
    Mar 23, 2021 at 8:53


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Browse other questions tagged or ask your own question.