Along this same vein of predictive anomaly detection, here are two videos (one from Netflix and one from Twitter) from the SF Metrics Meetup:

http://blog.librato.com/posts/sf-metrics-meetup-anomaly-detection



On 4 Mar 2015, at 06:26, Heiko W.Rupp <hrupp@redhat.com> wrote:


Am 25.02.2015 um 22:23 schrieb Matt Wringe <mwringe@redhat.com>:

With more historical data, it could get more interesting than a just an estimated straight line.

Absolutely. I did just send this screenshot to show that this is not a crazy idea but in fact already in production in consumer devices.


Although, bringing in historical data analysis and estimated ranges tend to make things more complicated and the charts more confusing.

We will need to use more complex historical data.
E.g. implement Holt-Winters forecasting.

 Heiko


_______________________________________________
hawkular-dev mailing list
hawkular-dev@lists.jboss.org
https://lists.jboss.org/mailman/listinfo/hawkular-dev