I have to say it’s refreshing and exciting to see data analytics garnering more attention this year at the National ACFE conference!
In multiple breakouts I was hearing the same terms over and over: “Hadoop” , “Big Data”, “Unstructured Data”, “Video Analytics”.  This is the cutting edge of fraud detection and I’m happy to see the ACFE embracing these topics.
Fraud analytics are not going away anytime soon.  For as much as was mentioned at the June 2014 conference there are many other topics to be covered.  From other analytic tools like R and Python, the use of Application Programming Interfaces (API’s) to analyze social media outlets like Twitter, Facebook and YouTube and diving even deeper into the mathematical theories used to detect fraud (beyond Benfords Law and into other mathematical models such as Levenshtein Distance, machine learning, etc).
For those interested in learning more in these areas I’ve put together some resources you can check out:
Websites:

• www.datasciencecentral.com
• www.analyticbridge.com
• www.informs.org
• www.wikipedia.org

Free Analytic Tools:

• R http://www.r-project.org
• SQL Server Express http://www.microsoft.com/web/platform/database.aspx
• Python https://www.python.org

Twitter:

• Cynthia Hetherington @HetheringtonGrp
• Informs @Informs
• Big Data Science @analyticbridge
• Data.Gov @usdatagov

…..and if you go through all the trouble to add those to your Twitter feed then add one more just for fun…..@jimmyfallon

So, everyone is using SQL, R, and SAS to analyze data to detect fraud.  What’s next?
There’s a great @analyticbridge blog post by Vishal Sharma about video analytics and it doesn’t take much vision to see its application in the area of fraud detection.  Think about the London bombings in 2005 for which surveillance cameras provided vital breakthroughs in the hunt for the bombers.  Now, combine this kind of surveillance footage with some advanced analytics and imagine software that is able to ‘understand’ when a crime is occurring and call police.
Vishal discusses four capabilities that sound like invaluable tools for fraud detection:

• Motion Detection:  Used to determine the presence of observed motion in the scene
• Object Detection: Used to determine the presence of a specific object or entity (a car, a person, etc)
• Video Tracking:  Used to determine the location of an object or person in the scene, sometimes in reference to another fixed point
• Pattern Recognition:  Software can be programmed to identify objects and should they change or move in a specific way (i.e., go missing or new objects appear) then the proper authorities could be alerted

For example, regarding pattern recognition, imagine a retail store being able to identify when a loading dock door is opened AND inventory is then loaded into a personal vehicle.  Video analytic software could pick up on this ‘red flag’ and immediately alert authorities.  Traditional data analytic tools (SQL, SAS, etc) might have been able to identify this theft…….days or weeks later.
Are you using video analytics?  Leave a comment below!