An Intensified Social Spider Optimization (ISSO) based Progressive Kernel Ridge Regression (PKRR) Classification Model for Automobile Insurance Fraud Detection

Abstract

Automobile insurance fraud detection is one of the most essential and highly demanding process need to be accomplished in the insurance industries. The existing works are highly concentrated on developing the fraud detection system by using various detection with the major issues of high complexity in computational operations, requires more time consumption, reduced speed of processing, and high error outputs. The proposed work objects to implement an efficient optimization based classification model for designing the automobile insurance fraud detection system. For this purpose, an Intensified Social Spider Optimization (ISSO) based Progressive Kernel Ridge Regression (PKKR) classification mechanism is developed in this work, which helps to accurately detect the insurance frauds from the given datasets. Initially, the data preprocessing and clustering operations have been performed by using the Distance based Fuzzy Clustering (DFC) approach, which produces the normalized clustered dataset for further processing. Then, the ISSO technique is deployed to select the best features for training the model of classifier based on the optimal fitness value. By using these features, the PKKR classifier accurately predicts the insurance frauds with reduced computational complexity and time consumption. During performance analysis, various evaluation indicators are used to validate the results of proposed detection mechanism.