Anomaly Detection using Similarity Approach on Airline Data

Abstract

Anomaly detection is to identify abnormal items, events or observations from the majority of the data. We applied similarity approaches to identify the abnormal observations from the Airline Data on chargeable weight. Chargeable weight is what the airline uses to determine the cost of the shipment. It may be either volumetric weight or gross weight, whichever is greater. Similarity approaches are applied to identify the abnormal observations on chargeable weight and evaluated the systems with the airline data. The precision, recall and F-measure values of the best system are 41.12%, 54.91% and 47.02% respectively.