Machine Learning is used to Detect Fraud in Insurance Claims

Abstract

Since a few years ago, an insurance company operating as a business has encountered fraud instances involving various kinds of claims. Because the amount fraudulently claimed is so large and could result in serious issues, various organisations are working with the government to identify and curtail these actions. Such frauds occurred in every area of insurance claim with high severity, for example, insurance claim towards the auto sector is fraud that is frequently claimed and prominent kind, which may be done by false accident claim. Therefore, our goal is to create a project that analyses set of insurance claim data to find fraud and inflated claims. The study uses machine intelligence algorithms to create a claim assessment and labelling model.
Additionally, a matrix of confusion comparison of complete machine intelligence mathods for categorization should be done in terms of soft accuracy, precision, recall, etc. Machine learning model is constructed for fraudulent transaction validation using PySpark Python Library.
According to estimates, fraud costs the insurance industry billions of dollars yearly and is on the rise across all industries. Insurance fraud is unlawful behaviour that is done with the intent to make money. Currently, this will be the most critical problem that many insurance firms throughout the world are dealing with. The primary factor has typically been recognized as one or more gaps in the investigation of bogus claims. Insurance fraud is a dishonest act that is frequently carried out with the intention of making money.
Every year, these erroneous claims cost the insurance sector billions in needless expenditures. The desire to deploy computer solutions to stop the growth of fraudulent activities , providing clients with not only a dependable and stable environment but also significantly reduced fraud claims.