Let's assume I have a table with columns datetime and volume for the search query "instagram".
Is there any ready python modules that would spot time interval of exponential growth?
Asked
Active
Viewed 173 times
0
Spektre
- 49,595
- 11
- 110
- 380
Baiel Muzuraimov
- 11
- 2
-
Fourrier transform of your time series can give you the frequencies and the period (the time interval you're looking for) is equal to 1/freq. I'm sure there are plenty of python module for fourrier transform. – Pierre Baret Jun 21 '22 at 11:08
-
why not code it yourself? see [estimate complexity from time measurements](https://stackoverflow.com/a/64784308/2521214) ... so you pick part of the graph and check which complexity it has ... to pick parts you can use crossing of some y value or FFT to determine offset and frequency ... then chose half periods ... If you want really just exponential growth then just test few "consequent" (with some gap) values against it and check whole interval something like sliding average where you move the N consequent points positions along the curve and check ... – Spektre Jun 22 '22 at 06:16