-
Couldn't load subscription status.
- Fork 337
-
I am trying to find this pattern in my timeseries data which is aggregated and essentially noise free.
image
Is stumpy's Matrix profile useful in identifying the event windows? It can be that the signal spends longer time at 100 (in the graph above) that might create a broader window.
Beta Was this translation helpful? Give feedback.
All reactions
@aeroaks Thank you for your question and welcome to the STUMPY community. For nearly exact pattern matches (where you know the pattern ahead of time), I recommend taking a look at this tutorial. Instead of computing a matrix profile, you compute something called a "distance profile" instead. Having said that, your window size seems a bit short and so you may have better luck with a different approach.
Replies: 1 comment 1 reply
-
@aeroaks Thank you for your question and welcome to the STUMPY community. For nearly exact pattern matches (where you know the pattern ahead of time), I recommend taking a look at this tutorial. Instead of computing a matrix profile, you compute something called a "distance profile" instead. Having said that, your window size seems a bit short and so you may have better luck with a different approach.
Beta Was this translation helpful? Give feedback.
All reactions
-
Thanks for the quick reply. Indeed, the type of data in my timeseries present opportunities to try more simpler analysis. I have given scipy.signal.correlate a try and it seems to perform good enough. I will look into the above mentioned tutorial too.
Beta Was this translation helpful? Give feedback.