Hi Jeyaram,

The 'logic' is difficult to explain as this is quite sophisticated.  And the 'logic' could be even different depending on the behavior of the metric itself.  If it was a moving average, it would be easy for me to answer your question, but it is a whole lot more than that (read: the crew developing this have tried to explain it to me and they lost me by the end of their second sentence 😉 ).

And note that we keep on refining this capability, we are aiming at making DataMiner better and more intelligent with every release.  For the forecasting, the most recent release now also looks at the time frame you are looking at, because forecasting a metric for the next hour is different as compared to the next day, next week, next month.

And that's also the case for all of our other Augmented Operation capabilities, such as proactive alarming, change event and anomaly detection, automatic tagging, incident identification, etc.

By the way if you could share some screen captures of some nice forecasts with us, that would always be appreciated.  You can add them in the questions here.  Our analytics team would love to see those and learn from what happens also in the field.

Cheers!