Slide 1

Anomaly or outlier detection involves identifying elements which do not conform to a specific expected pattern. Typically, anomaly detection analyses are undertaken to identify exceptions like fraud, defects, errors, medical ailments, security hazards, etc.

The KMS Analytics practice applies a combination of Advanced Analytics algorithms and statistical softwares, Business Intelligence and Big Data analytics (based on the data complexity) to analyze, visualize and highlight anomalies in data. Based on these insights, organizations can take proactive actions towards early mitigation of threats / risks that might be posed by such cases.

We utilize multiple anomaly detection techniques which include Statistical Model based (anomalies are identified as those elements which do not fit the model), Proximity based (anomalies are "distant" from most other elements), Density based (Anomalies are in the low density areas and are far from others in the regions of higher densities) or Cluster based (anomalies are small clusters that are far from most other clusters).

Our methodologies can be applied in diverse areas like:

Procurement Analytics: Pre-contract and post-contract procurement frauds.
Credit Card Analytics: Anomalous transactions in credit cards.
Telecom Analytics: Telecom frauds related to subscription, internal, roaming, etc.
Healthcare Analytics: Anomalies in patient vital signs using signal processing and threshold criteria.
Transportation Analytics: Wrong fare occurrences in smart transportation systems, including device or process related faults.
Security Analytics: Network intrusions or hacker attacks, or internal security hazards.
Machine Data Analytics: Device related faults and predictive maintenance.
Sensor Analytics: Anomalies in sensor data to provide alert mechanisms.
Network and Web Analytics: Anomalies in network or website traffic that can impact business.