App abnormal events

The App abnormal events anomaly insight lets you know when issues are detected in newer versions of an app, but those same issues may not be as prevalent in previous versions of the same app.

For example, if your fleet has multiple versions of the Samsung Knox Capture app deployed, Anomaly detection can let you know if a newer version of the app encounters more or fewer issues compared to previous versions.

How it works

Using an ANOVA (Analysis of variance) data model, Anomaly detection analyzes your fleet’s app-related activity and checks for all abnormal app issues reported over the last 60 days. Once a week, an anomaly report is generated, and the results are displayed on the Anomaly detection panel.

If a newer version of an app reports significantly more or fewer abnormal events than a previous version, then an Issue or Improvement is reported respectively.

Using the previous example above, if you deployed v1.50 and v1.49 of Samsung Knox Capture in your fleet, and v1.50 reports 11% more abnormal events than v1.49, then Anomaly detection reports this as an Issue. You’d see the following sample information on the dashboard:

Issue found on 9.0% of devices

Detected 11.0% more abnormal app events on v1.50 of the Samsung Knox Capture app compared to v1.49.

This information lets you know that 9% of all devices in the fleet have abnormal events detected in v1.50 of the app over the last 60 days, and that v1.50 is less stable than v1.49 since it reports an average of 11% more issues per device per day.

If v1.50 reports fewer anomalies than v1.49, then Anomaly detection reports this as an Improvement, and you’d see the following sample information:

Improvement found on 9.0% of devices

Detected 11.0% fewer abnormal app events on v1.51 of the Samsung Knox Capture app compared to v1.50.