The Fifth Elephant 2013

An Event on Big Data and Cloud Computing

Insurance Fraud Modeling & Business Intelligence Framework

Submitted by Dr.S.Jayaprakash,Ph.D (@drjayaprakash) on Tuesday, 21 May 2013

Section: Product Demos Technical level: Advanced


We have successfully created a product on Insurance Fraud Modeling Framework backed by the robust Business Intelligence Analytics. Some of the USPs of this model are (a) Quick to Deploy within few weeks (b) Proven Statistical Models (c) Minimum of 10X ROI for Insurance companies (d) Deployment made possible at the fraction of budgets of IT Departments. The principles behind the framework is scalable across various industries not limiting to Insurance alone.


Frauds is rampant in every sector. Insurance is one of the industries more prone to fraud. It has been always a challenge the insurer to trap the frauds due to the cost involved, complexity of factors etc. This framework has overcome the hurdles.

Speaker bio

In Insurance sector, I come with the rare combination of exposure of Data Warehousing Technology, Statistical Techniques and a Ph.D degree in Insurance & Risk Management area. Also, I worked in the Director-Cadre at Metlife managing the Fraudulent activities which has given me practical hands on experience. This framework is a combination of various experience and this is a low cost yet high beneficial model. Another USP of the model is that this can be extended to other industries as well. i.e. the principles.


  • Joydeep Sen Sarma (@jsensarma) 6 years ago

    Seems promising. Can you provide some technical details?

  • Dr.S.Jayaprakash,Ph.D (@drjayaprakash) Proposer 6 years ago

    Identifying frauds is definitely not like a “TurnKey” solution. Before the actual big data that we are talking about, the real big data concept emanates from Fraud analytics, wherein we do all kinds of analysis across all medium. What we have created is a robust business intelligence framework, wherein each record will be flagged across 10-15 dimensions. So, by reading a record and its associated flags, the story around the transactions can be told. On top of this flags, statistical models are built based on techniques like Principal Component analysis, Logistic Regression, Associative Rules, Self Organizing Maps, Naive Bayes etc. to predict the outcome.
    BusinessIntelligence Framework has been built upon a platform that has the potential for seamless integration with various data sources and business intelligence can be built upon within a few days based on the fraud rules.
    Though this solution has been built upon insurance models, we realized later that the rules are industry agnostic and can be applied across various industries.

Login with Twitter or Google to leave a comment