Industry 4.0, AI, and Data Science

Research Trends and Challenges

Vikram Bali, Kakoli Banerjee, Narendra Kumar, Sanjay Gour, Sunil Kumar Chawla, Vikram Bali, Kakoli Banerjee, Narendra Kumar, Sanjay Gour, Sunil Kumar Chawla

• Publisher: CRC Press
• Year: 2021
• Print ISBN:
  9780367562854
• eBook ISBN:
  9781000413489
• Topic: Technology & Engineering
• Subtopic: Cyber Security
• Index: Technology & Engineering

The aim of this book is to provide insight into Data Science and Artificial Learning Techniques based on Industry 4.0, conveys how Machine Learning & Data Science are becoming an essential part of industrial and academic research. Varying from healthcare to social networking and everywhere hybrid models for Data Science, Al, and Machine Learning are being used. The book describes different theoretical and practical aspects and highlights how new systems are being developed.

Along with focusing on the research trends, challenges and future of AI in Data Science, the book explores the potential for integration of advanced AI algorithms, addresses the challenges of Data Science for Industry 4.0, covers different security issues, includes qualitative and quantitative research, and offers case studies with working models. This book also provides an overview of AI and Data Science algorithms for readers who do not have a strong mathematical background.

Undergraduates, postgraduates, academicians, researchers, and industry professionals will benefit from this book and use it as a guide.

1

Predicting Fraudulent Motor Vehicle Insurance Claims Using Data Mining Model

Jacob Muchuchuti and Stewart Muchuchuti

Contents

1.1Introduction

1.2Proposed Model

1.2.1Statement of the Problem

1.3Literature Review

1.3.1The Concept of Insurance

1.3.2Insurance Fraud

1.3.3Fraud Predictive Variables

1.3.4Machine Learning Algorithms

1.3.4.1Naive Bayes

1.3.4.2Decision Trees

1.3.4.3Logistic Regression

1.3.4.4Support Vector Machines (SVM)

1.4Methodology

1.4.1Data Acquisition and Description

1.4.2Data Pre-processing

1.4.3Encoding

1.5Application of Classification Algorithms

1.6Results

1.6.1Attributes of Predictor Variables

1.6.2Variables with the Most Predictive Influence

1.6.3The Classification Algorithms Used

1.7Conclusions and Future Scope

1.7.1Recommendation of Future Studies

References

1.1Introduction

Insurance, as a risk transfer mechanism, has become the hope for individuals, social groups, and businesses (Viaene and Dedene 2004). The benefits of insurance therefore may not be underestimated to insurance companies, policyholders, and the economy at large. However, players in the insurance industry are faced with a challenge that undermines the gains that may be realized from the use of insurance as a risk management concept as insurance fraud is on the increase hence depleting the funds paid in by the many honest customers to cover genuine losses (Insurance Europe 2013). While insurance companies have tried to develop effective procedures for identifying, investigating, and deterring fraudulent activities, according to Mosley and Kucera (2014), these initiatives can go that far. However, despite the experience that the professionals who are involved in the management of fraudulent claims may have, there are not enough of these trained eyes to review every claim, resulting in some fraudulent claims ending up slipping through the cracks, resulting in payments being made that were not supposed to be processed.

1.2Proposed Model

The research sought to develop an input/output (black box) data mining model that aimed at assisting insurance companies to be able to predict fraudulent claims in motor vehicle insurance using observable attributes, using available algorithms.

1.2.1Statement of the Problem

According to Chudgar and Asthana (2013), there has been an increase in fraudulent insurance cases lately and the trend has been observed from an international, continental as well as regional perspectives. The United Kingdom government estimates that the insurance industry faces about £3.4 billion of detected and undetected insurance fraud (HM Treasury 2015). In the United States of America, the Coalition against Insurance Fraud (2012) posits that insurance fraud is ranked second in the list of expensive crimes with 10 percent of the fraudulent transactions being from fraudulent claims (Insurance Information Institute 2019). Ernst and Young (2011) observed in their research that, out of the total claims that are presented in health insurance in India, nearly 25 percent are fraudulent. Africa has not been spared with the trend of fraudulent insurance claims. South Africa, being the continent’s economic hub, estimates that, out of ZAR45 billion spent on claims in South Africa, about ZAR5.5 billion is lost annually through insurance fraud with 32 percent of the latter being suspected to be fraudulent claims (Risk & Insurance Zimbabwe 2018). These fraudulent claims are driving up insurance companies’ overall cost resulting in their continued existence being threatened. Given the above statistics, the researchers’ view is that insurance fraud is an economic cancer that has ravaged economies across the entire globe. We are of the view therefore that the use of predictive analytic tools such as data mining and machine learning become handy in dealing with the challenge at hand.

1.3Literature Review

1.3.1The Concept of Insurance

Insurance may be considered to be as old as humanity although the way it was being used then and how it is done today may be different. There are two major types of insurance policies: life insurance that normally have a longer term to maturity that is more than a year and the short term ones that are renewed on an annual basis. Automobile insurance or motor vehicle insurance, which is the research fulcrum, belongs to the latter. Auto insurance is a contract between a policyholder (insured) and the insurance company (insurer) that protects you against financial loss in the event of an accident, or theft of your motor vehicle as well as fire (www.iii.org). The insurance company should therefore restore the insured to the same financial position he/she had before the occurrence of the loss The insurance company agrees to pay the insured’s losses as outlined in your policy and auto insurance normally provides coverage for property in cases such as damage to or theft of the vehicle; liability, where an insured is legal responsible to others for bodily injury or property damage; or medical, being the cost of treating injuries, rehabilitation, and sometimes lost wages and funeral expenses for the dead.

1.3.2Insurance Fraud

The Insurance and Pensions Commission of Zimbabwe (IPEC) on its official website (www.ipec.co.zw) defines insurance fraud as cases in which individuals or entities lie to an insurance company for the sake of getting financial compensation for something they would not have received had the truth been told. The definition is supported by the Institute of Internal Auditor’s International Professional Practices Framework (IPPF) as quoted by Ernst & Young (2011) that defines fraud as any illegal act characterized by deceit, concealment, or violation of trust. As mentioned earlier, insurance is based on the principle of mutual benefit and is designed to protect against significant, but uncertain losses; however, fraudulent claims undermines this system since they deplete the funds paid in by the many honest customers to cover genuine losses (Insurance Europe 2013).

Insurance fraud has devastating effects to all the insurance industry players, including and not only limited to insurance companies and their customers. Insurance fraud thus has adverse effects to the efficient functioning of a national economy depending on the magnitude of fraud. One major impact of insurance fraud to the insurance companies is that it drives up the costs of doing business and may lead to huge losses to be incurred. These costs are in the form of investigation of all the claims as well as the ultimate payment of claims that are not genuine.

It is without doubt that the insurance expense of the average household would also increase due to higher premiums that are then paid in order to cover the cost of the fraudulent transaction. This has the effect that policyholders no longer pay fair premiums and that has the resultant effect of them shunning the insurance market. This emanates from the concept of pooling where policyholders share losses.

According to Mpofu, De Beer, Nortje, and De Venter (2010), fraudulent claims have a negative impact on policyholders due to delayed processing of genuine claims, resulting in policyholders failing to get value for their money.

1.3.3Fraud Predictive Variables

The variables, also known as attributes, are used to organize records of data in database tables. Hargreaves and Singhania (2015) in their research on fraudulent insurance g...

Table of contents

• Cover
• Half Title
• Series Information
• Title Page
• Copyright Page
• Contents
• List of contributors
• Preface
• Editors
• 1 Predicting Fraudulent Motor Vehicle Insurance Claims Using Data Mining Model
• 2 Novel 8: 1 Multiplexer for Low Power and Area Efficient Design in Industry 4.0:
• 3 Data Science and AI for E-Governance: A Step towards Society 5.0
• 4 Application Areas of Data Science and AI for Improved Society 5.0 Era
• 5 Applying Machine-Learning and Internet of Things in Healthcare
• 6 Artificial Intelligence: The New Expert in Medical Treatment
• 7 Machine Learning Approach for Breast Cancer Early Diagnosis
• 8 Intelligent Surveillance System Using Machine Learning
• 9 Cyber Security: An Approach to Secure IoT from Cyber Attacks Using Deep Learning
• 10 Learning the Dynamic Change of User Interests from Noise Web Data
• 11 Artificial Intelligence Techniques Based Routing Protocols in VANETs: A Review
• 12 A Comparison of Different Consensus Protocols: The Backbone of the Blockchain Technology
• 13 Blockchain in AI: Review of Decentralized Smart System
• 14 Financial Portfolio Optimization: An AI Based Decision-Making Approach
• 15 Intelligent Framework and Metrics for Assessment of Smart Cities
• Index