Anomaly detection in a network based on a key performance indicator prediction model
摘要
說明書
In some implementations, the network monitoring platform may update the prediction accuracy value by determining an average of the amount of error in the prediction of the particular value of the KPI and amounts of respective errors in predictions of historical values of the KPI. The average may be a weighted average, with the weighting based on respective ages of the predictions of the historical values of the KPI (e.g., amounts of errors of older predictions may be weighted to have less of an effect on the prediction accuracy value than recent predictions). In some implementations, the amounts of errors of predictions older than a threshold are not used to determine the prediction accuracy value.
In some implementations, the network monitoring platform may update the prediction accuracy value periodically, upon identification of a quantity of KPI values as anomalous or not anomalous, by request from the network operator, and/or the like.
