0

I've got the following set of data : enter image description here

As you can see, I've got two plateau at 1987 and 3757. The problem is I can't find a good algorithm in order to extract them. I've already tried this algo, without any success.

Any help will be welcomed.

Regards

ravenspoint
  • 19,093
  • 6
  • 57
  • 103
Khan From Chedum
  • 21
  • 1
  • 6
  • 3
    It is actually hard to see that there is a **plateau** there (assuming there is also not one close to 1102). Have you tried low-pass filtering your data? If it is not real-time, you can try a simple moving average. Without the data it is difficult to find the correct parameters, but it looks like something between 30 to 50 samples will help remove the noise. – nonDucor Jun 26 '22 at 14:14
  • Thank you for your response. It's not real time data, so I'll try to make a moving average on the data. – Khan From Chedum Jun 26 '22 at 14:32
  • 1
    If it's not real-time you can simply test for structural breaks. See e.g. this question with accompanying answers: https://stats.stackexchange.com/questions/395078/how-to-detect-and-quantify-a-structural-break-in-time-series-r – Jean-Paul Jul 26 '22 at 09:25

0 Answers0