Adaptive model charts

The Adaptive alert's page includes the adaptive model chart, which makes it easier to observe traffic rates and note when the threshold was reached or crossed.

You can use it to analyze historical traffic data. Its time line can be simply changed by either clicking appropriate buttons at the top (four hours (4h), one day (1d), one week (1w), two weeks (2w), and one month (1m)) or dragging the handles on the miniature graph.

There are three things about the charts that you can observe:

  • Blank periods. If data can’t be pulled from a system, the Alerts application leaves it missing rather than extrapolate using incomplete information. These missing periods are shown as blank on the graph.
  • No predictions. For the chart to show predictions, there must be three hours' worth of data. Note that it also depends on the predictive model selected. For Weekly, you need to have data from at least one week and one day for Daily for the alert to monitor the traffic. If you have the necessary data and you don't see the predictions, then check whether you didn’t set the thresholds to None.
  • Wrong predictions after a major event. The traffic during a major event changes future predictions. It’s enough to wait for the traffic to readjust back down to normal levels.

What's also important about the charts is that the data presented on those charts can differ from reporting data. It's because the system uses traffic data from an internal data aggregation system. Data is gathered in real time and published when most systems report. It's possible that traffic may differ by up to 10% compared to billing data, if some systems fail to report in real time. Billing data is always more accurate since it's not generated until all systems report traffic.