PREDICTIVE ANALYTICS IN FRAUD Click Scott to White edit Master subtitle style Business Development Manager
Why predict? Organizations that use predictive business performance metrics will increase their profitability by 20% by 2017. Gartner Press Release, Gartner Says Organizations Using Predictive Business Performance Metrics Will Increase Their Profitability 20 Percent by 2017. January 16, 2014. http://www.gartner.com/newsroom/id/2650815
Challenges in Deploying Predictive Analytics Despite the buzz, the percentage of organizations that have implemented predictive analytics has remained surprisingly flat. Wayne Eckerson, Principal Consultant Eckerson Group Challenges in Deploying Predictive Analytics TechTarget, Making Predictive Analytics Pervasive, by Wayne Eckerson (May 2014) Copyright 2015 Predixion Software. All rights reserved.
What We Do Predixion provides real-time, predictive analytics at the decision point to improve outcomes. Analyze & predict everything Take action Build Compare Visualize BI tools Apps Shape Collaborate Data from anywhere Combine Deploy anywhere Cloud Services Devices Machines Streaming Analytics (CEP) Copyright 2015 Predixion Software. All rights reserved.
How Predixion is Unique T H E B E S T O F B O T H DATA SCIENCE TOOLS Designed for Data Scientists Solves multiple problems for data scientist Sophisticated, complex & time-consuming Difficult for the business to see value Provides value to both the business and the data scientist Business realizes immediate value in quick to implement applications that improve decisions & outcomes Data Scientist benefits from an easy to use, yet powerful platform that: Leverages existing algorithms, packages and models Demonstrates their value to the business quickly Solves multiple use cases; not a one-off niche application Empowers analysts with predictive capabilities BLACK BOX APPLICATIONS Designed for business users Solves a single, niche problem using predictive modeling Business gets immediate value for solving that one problem Difficult for data scientist to see value
Predictive Use Cases with Impact
Predict & Intervene on Insurance Claims Fraud THE CHALLENGE PREDIXION SOLUTION BENEFITS $80B in fraudulent claims per year in U.S. 1 Fraud is on the rise in all markets, especially in insurance & banking Manually fighting fraud is expensive, contributing to higher premiums Effective anti-fraud programs include predictive modeling 2 Uses multiple enterprise data sources claims and demographic data for highest accuracy Predictive model built using Predixion Insight Runs model in real-time to risk score claims within segmented profiles Displays scores & interventions in custom portals or pushes to existing LOB systems at the decision point Empowers business users within existing systems with predictive analytics Multiplies the effectiveness of existing staff by greatly increasing the speed of fraud detection Reduces costs by avoiding fraudulent payouts 1 Source: The Impact of Insurance Fraud bycoalition Against Insurance Fraud 2 Source: Insurance Claims Fraud: Assembling An Analytics Tool Box by Michael Costonis, Executive Director of Insurance Practice at Accenture. Copyright 2015 Predixion Software. All rights reserved.
Case Study: Quickly Detect Costly Fraudulent Claims CLIENT: A national healthcare insurance provider PROJECT GOAL: Replace a manual, labor-intensive document review process for identifying errors in claims with an accurate, automated approach RESULT: Implemented in only 4 weeks Reduced contract auditing time by 95% Identified $5M in erroneous claims almost immediately Copyright 2015 Predixion Software. All rights reserved.
EBT Fraud Background Electron Benefit Transfer (EBT) Card replacing all paper coupons and checks (food stamps) Federally funded program with partnerships with state agencies Food and Nutrition Service (FNS) has a dedicated team of over 100 investigators nationwide Supplemental Nutrition Assistance Program (SNAP) is the federal sponsored program The Food and Nutrition Service works with State agencies, nutrition educators, and neighborhood and faithbased organizations to ensure that those eligible for nutrition assistance can make informed decisions about applying for the program and can access benefits 47 million participants with $70 billion/year in benefits distributed EBT is the technology and card services leveraged to deliver benefits to participants in SNAP Qualified applicants submit an application to get approved SNAP benefits Average household benefit is $500/month The most common fraud abuse averages $20,000/month per case Federal government tracks $1 billion a year in known fraud; undetected fraud is estimated 5-10 times higher Local state agencies are limited with few resources to investigate fraud (SNAP is Fed. Program) Great opportunity for predictive analytics to improve the process by surfacing high risk cases for investigation http://www.fns.usda.gov/snap/supplemental-nutrition-assistance-program-snap http://www.fns.usda.gov/snap/supplemental-nutrition-assistance-program-snap
Role of Predictive Analytics in EBT Fraud Incorporating current business strategies to combat EBT Fraud with predictive analytics seems to be the most effective approach to reducing fraud in SNAP State agencies can save millions in taxpayer dollars annually by integrating data from multiple data sources and leveraging predictive analytics Tying both the investigation processes directly with predictive insight maximizes resources utilization and improves the accuracy in identifying high risk fraud early SNAP EBT purchase transactions help to identify suspicious transaction patterns Detect suspicious EBT fraud activity in real-time, prioritizes and assign cases automatically to investigators to detect the fraud early in its life cycle. Route the predictive insight and alerts to the appropriate decision makers in the field Utilize Predixion s machine learning capabilities to surface relationships between participants and retailers engaged in fraud to facilitate the investigation, capture and display of key information pertinent to the case.
Landing Screen
Program Manager Screen
Program Analyst Screen
Program Investigator Screen
Program Participants Trends
Program Participants Specific Case Drill-in
Program Actions Screen
Program Actions Specific Case Drill-in
Thank you Scott White swhite@predixionsoftware.com