DataSentics

Anomaly detection

Introduction

The success of the brand depends on customers’ satisfaction with offered products or services. But in today’s world, the internet connection becomes even more crucial.

Business case

Since services such as internet connection are provided round-the-clock, effective anomaly detection would require 24/7 monitoring. Trained employees would need to detect the outages before the first displeased customer’s call.

Solution

Anomaly detection algorithms can achieve high accuracy in detecting data anomalies immediately as the data is available. Moreover, the algorithms can generate alerts for further investigations of the data, and even classify the anomaly types and offer more insights into what happened. This is where using machine learning can help immensely.

DataSentics offers expertise in various types of anomalies, with references, e.g., in communications of IoT devices connected to a TV, transmitted by wireless mobile networks. Outages detected in this case were related to power malfunctions in cell towers due to bad weather, human errors, or even floods.

Benefits

  • Early detection of an outage
  • More in-depth analysis of the network outages in general
  • More insights into possible future changes to the network due to collected anomalous data
  • Identifying unusual invoices, invoices from uncommon accounts, unusual amounts, or at strange dates with the help of AD
  • Replacing error-prone manual work and repetitive, mundane tasks
  • Significant cost savings