Selecting key performance indicators for anomaly detection analytics
US10169731B2 · kind B2 · utility
Assignee
Inventors
Key dates
| Filing date | Nov 2, 2015 |
| Grant date | Jan 1, 2019 |
| Priority date | — |
| Expiry date | Mar 6, 2037 |
Classification
- Technology area (CPC H)Electricity
- CPC primaryH04L67/564
- WIPO fieldDigital communication
- WIPO sectorElectrical engineering
Abstract
A computer program product is provided and includes a storage medium having program instructions. The program instructions are readable and executable by a processing circuit to cause the processing circuit to determine from historical data which anomaly detectors are associated with key performance indicators (KPIs), to extract descriptors of the first metric group from the KPIs to create a first feature profile thereof, to repeat the determining and the extracting with respect to historical data of second and third metric groups to create second and third feature profiles thereof, respectively, to ascertain which of the second and third feature profiles has a greater correlation to the first feature profile and to rank the second and third feature profiles based on which one has the greater correlation and compare the higher ranked one to future metrics in current or subsequent environments to determine if specific KPIs are likely to produce anomalies.
Source: USPTO / EPO open patent data. Objective bibliographic and citation counts.