Im trying to find a method to filter data sets that looks almost like this
- The blue is the original signal
- The orange is filtered using average of batches
- the green is the peaks using the
scipy.find_peaks
Some backgroud: The graph represents weight for two periods, i would like to find the circled points so i can tell in the first period the collected weight was X and in the second period it was Y.
The problem is my find_peaks paramters arent applicaple on each data set, so everytime i have to customize its parameters which dosent make any sense
I tried to calculate the slope but that didnt work out well.
Which method or algorithm would you use for such a problem?