Fleet Management News & Business Info | Commercial Carrier Journal
Issue link: http://read.dmtmag.com/i/716013
COMMERCIAL CARRIER JOURNAL | AUGUST 2016 83 How machine learning detects waste, fraud, maintenance needs BY AARON HUFF M achine learning is in many of the applications we use daily. Netfl ix uses it to recommend mov- ies, Facebook to tag our friends in photos, and Google and Microsoft to automatically sort our unwanted email to the junk folder. The term describes methods that use statistical models, real-time analysis and algorithms to fi nd patterns in data and deliver instant recommendations. Machine learning also is present in a number of fl eet management systems, particularly those that process a large volume of transactions such as fuel purchases. Fleets have used electronic payment systems for more than 30 years to control the locations and gallon and dollar limits of their fuel transactions. However, these "card controls" do not always stop transactions that are wasteful or even fraudulent. Drivers might use company fuel cards to buy premium grade instead of regular gasoline, purchase at locations with higher prices or occasionally add gallons to their personal vehicles. Rather than dump fuel transaction data into spreadsheets for analysis to fi nd exceptions, fl eets now are able to obtain real-time alerts when algorithms detect data patterns that call for immediate action. Fueling big data Wex, a provider of fuel card and corporate payment sys- tems, formed its Wex ClearView analytics team last year. The company saw an opportunity to leverage its large volume of fuel transaction data to create new analytical products and services for customers. Wex captures millions of nationwide fuel transactions per week from its local and over-the-road fuel card programs. Local fl eets use the Wex Fleet card, while over-the-road cus- tomers use purchasing systems from company subsidiaries EFS and Fleet One. Wex ClearView developed machine-learning algorithms to identify exceptions and created an Exceptions Module in the Wex Online portal to share the results, says Kurt Thear- ling, the company's vice president of analytics. The Exceptions Module has a dashboard-style interface that displays the status of each exception using green, yellow and red indicators. Those exceptions include nonregular fuel purchases, odometer/mpg readings, fuel purchases exceeding vehicle tank capacity and more. Clicking on any of the indicators brings up a drill-down report of the drivers and transactions tied to the exceptions. Algorithms detect patterns and instances of fraud, such as when a driver enters incorrect odometer readings at the pump. Another fraud pattern is when the number of gallons purchased does not match the expected gallons consumed between fueling events based on the vehicle's average mpg and tank capacity. Wex ClearView also uses telematics data to detect fraud. The company has partnerships with telematics providers to access data with its customers' permission. One pattern that algorithms detect is when the GPS location of a service station does not match the location of a vehicle at the time of a fuel transaction. Including the location of a driver's home in the real-time analysis also is valuable. One fraud pattern is when a driver