I did some researches on SO and found many related questions/answers about pattern recognition. I was hesitating whether to post this question because this could be duplicated... But since I'm a noob in this research area and there're too much information overwhelming my head, I'm worry about that I still can't properly solve my problem after learning those concepts. So I decide to ask, probably again.
I have such a plot of data
The blue part gives an image of how normal data should change. But that is not the interesting point. After that, the data should remain constant, steady to a specific value. However, because of exogenic factors, there are jitters occasionally appearing there.
I want to know when such kind of jitters occur, and how many of them. A possible result to the plot of data could be: t1_start, t1_end, with 5 jitters, and t2_start, t2_end, with 5 jitters.
The shown case can be a simplified version. The jitters can occur on dynamic data with different amplitudes. I also want to know if this belongs to the fields like pattern recognition, signal processing, etc, with those knowledge one is possible to come up with solutions.