How can I recognize patterns with different sizes in a time-series? Imagine that I have a template pattern and I need to find that pattern in a symbolic representation of the time-series (stock market prices). However, the pattern can appear with different lengths. I found dynamic time warping here, for example. Is this a possible solution?

  • $\begingroup$ a naive solution would be to generate several templates in different scales, and convolve with the time series several times. $\endgroup$ – yoki Dec 11 '14 at 12:35
  • $\begingroup$ What is a symbolic representation of the time-series? $\endgroup$ – mpiktas Dec 11 '14 at 12:56
  • $\begingroup$ The naive solution is my current idea. When I say symbolic representation, something like taking the time derivative of the time-series and mapping the value to a given symbol.Thanks for the help. $\endgroup$ – rfty Dec 11 '14 at 13:33
  • $\begingroup$ There are no patterns in stock markets. $\endgroup$ – Aksakal Dec 11 '14 at 14:41

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.