alarms based on statistical analysis, or even deep learning #1490

Open
ktsaou opened this Issue Jan 1, 2017 · 0 comments

Projects

None yet

1 participant

@ktsaou
Member
ktsaou commented Jan 1, 2017

Happy New Year!

I found today the excellent article of @ZygmuntZ at: http://fastml.com/bayesian-machine-learning/

I have spend quite some time reading statistics and deep learning algorithms, trying to figure out if we can add a better / smarter alarming mechanism to http://my-netdata.io

But, I am not a data scientist, so most of it is still confusing for me (yeah, I know).

What I read though in such articles, is really interesting.

For example, if we could somehow know the probability of the value we just collected, based on the history of each metric we have, we could have a lot smarter alarms. Alarms that could help us pinpoint the conditions under which our web servers a facing abnormally high or low traffic, that our databases are struggling to serve the load offered to them, that our systems are unreasonably throttling requests, etc.

If you understand the math and you are willing to help thousands of people monitor their servers in a better way, please join the discussion here and help me figure this out...

Thanks!

@ktsaou ktsaou added the enhancement label Jan 1, 2017
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment