Use of Particle Swarm Optimization for Feature Selection and Data Mining Methods for Efficient Detection of Automobile Insurance Fraud (original) (raw)

2018 International Conference on Recent Innovations in Electrical, Electronics & Communication Engineering (ICRIEECE), 2018

Abstract

In this research, we have carried out a systematic study in automobile insurance fraud detection. The fraudster, their main types and subtypes of known insurance frauds has been defined. We have categorized, compared, and summarized from almost all published technical and review articles in this domain within the last 10 years. A novel scheme has been proposed that uses a Particle Swarm Optimization (PSO) based feature selection method for extracting irrelevant and redundant features in automobile insurance dataset. As the dataset is highly skewed in nature, we have devised a Quarter Sphere Support Vector Machine (QS-SVM) based under sampling approach for data balancing. Thereafter, we have employed Decision Tree (DT) and Logistic Regression (LR) for classification purpose on the balanced data. The effectiveness of our proposed methodology is evaluated experimentally using a real world automobile insurance fraud dataset taken from literature.

Suvasini Panigrahi hasn't uploaded this paper.

Let Suvasini know you want this paper to be uploaded.

Ask for this paper to be uploaded.