A Possible Framework For Big Data Analytics For Fraud Detection -In Vehicle Insurance

Abstract

Big Data Analytics Technology has been a growing industry in solving our today’s daily activity, meanwhile in enhancing the industry of insurance by detecting fraud in all claims of the client using information from customer driving history and social media status data. Fraud and any illegal activity are the main headache of the vehicle insurance industry that change its characteristics like virus depending on the technology that the fraudster ability of using it. So insurance companies must use an advanced technology that analysis the claim and predict the outcome for decisions. Traditional methods, Data Mining and expensive Algorithm of fraud detection use only stored data or structured data to analysis and identify claims but we have to use data from different sources, indifferent format i.e. Unstructured data for a better balanced decision. The designed Possible Big Data analytics framework for fraud detection in vehicle insurance has three main parts: Data Acquisition and Preparation: The data about the insured person vehicle will gathered from different sources, then all the data will be in to Big Data Integration-Hadoop that use: Crawler: Is a program that visits websites and reads their page to create entries for a search engine with having: Flume, Sqoop. Integration: Is combine data from disparate source in to meaningful and valuable information i.e. Integration Big Data Analytical State: Integrate-Customer Behavior, Drinking Habit, Bank Statement,, claim history, Social Media status:- FB, Twitter, Instagram, google+ and others to protect insurance and claim Patterns. Then finally Delivery/Visualization of results, i.e. Data exploration, Visualization using query language either Hive, R for Fraud detection. The proposed possible Big Data analytics framework for fraud detection in vehicle insurance will help Insurance companies to know all of the data to gain basic insights of the dataset and prepare analysis: called a descriptive analysis then predictive analysis based on results of dataset. So using different V’s of property of Big Data, different Analytics methods(descriptive and predictive) and with the help of Hadoop ecosystem, including R programming language we could to analysis datasets from different sources(Social media activity, telemetric software, personal profile and behavior and others ) in insurance companies.