I'm starting to learn machine learning, and have honed in on a problem to get my hands dirty with: how to detect a trend reversal in a time series.

Trend reversal example

I've asked this question on stats.SE more than a year ago, but now I'm looking for a machine learning approach. How would I go about this specific problem vs. generally identifying patterns in time-series data?

  • $\begingroup$ It looks like you already asked this question on Statistics.SE, and got an answer there: stats.stackexchange.com/q/239868/2921. Please do not post the same question on multiple sites. Each community should have an honest shot at answering without anybody's time being wasted. Also, if you already got an answer, you should describe the suggestions you've already got and explain whether they met your needs or not and why, so people don't waste their time telling you something you already know. $\endgroup$ – D.W. Jan 26 '18 at 1:28
  • $\begingroup$ I'm voting to close this question because it was cross-posted on Stats.SE. $\endgroup$ – D.W. Jan 26 '18 at 1:29
  • 2
    $\begingroup$ @D.W. - I had indeed posted the question on Cross Validated over a year ago, but now I'm looking for an ML-oriented approach. $\endgroup$ – Gascoyne Jan 30 '18 at 11:11
  • $\begingroup$ Why? What's wrong with the approaches you've already been given? If you've rejected the approaches suggested there I'm worried you might just reject any approach we suggest, too. If you rejected their approaches that's presumably because you have some requirement that they didn't meet. If so, please edit your question to state your requirements, and state what approaches you considered (and what approaches have been suggested to you) and why you rejected them. The question needs to specify how you plan to evaluate proposed solutions, so we can evaluate them too. $\endgroup$ – D.W. Jan 30 '18 at 18:09
  • $\begingroup$ Just re-posting the same question without indicating you've posted it before and gotten some other answers is problematic -- it risks people duplicating answers you've already gotten, which would waste your time and theirs. You should be summarizing and linking to all related resources you've found (including the prior question you asked) and giving a little overview of the subject and how they do/don't meet your needs. Try to make your question useful for others and show your work so far. See cs.stackexchange.com/help/how-to-ask. $\endgroup$ – D.W. Jan 30 '18 at 18:09