A Class Imbalance Learning Approach to Fraud Detection in Online Advertising

  • Baruhupolage Kasun Perera

Student thesis: Master's Thesis

Abstract

By diverting funds away from legitimate partners, "Click Fraud" represents a serious drain on advertising budgets and can seriously harm the viability of the internet advertising market. As such, fraud detection algorithms which can identify fraudulent behavior based on user click patterns are extremely valuable. This thesis investigates the problem of click fraud detection and the performance of several detection algorithms. These fraudulent behaviors exist in credit card transactions, insurance claims, business transactions and even new fields, such as smart grids. Fraud detection systems typically suffer an inherent problem called "Imbalance Class Distribution", and under this scenario conventional classification algorithms fail. In this thesis we propose a novel approach for click fraud detection, which is based on a combination of different models, such as Ensemble Learning and Sampling Methods. The proposed models are evaluated in terms of the resulting precision, recall and the F-Measure. The final method, based on four different learning algorithms, proved to be stable with respect to all performance indicators. It showed superior results on training, validation and test datasets, thus demonstrating its generalizability to different datasets.
Date of AwardJun 2013
Original languageAmerican English
SupervisorU Zeyar Aung (Supervisor)

Keywords

  • Optical Data Processing; "Click Fraud"; Fraud Detection; Online Advertising; Feature Engineering
  • Imbalanced Data.

Cite this

'