Machine learning describes methods that use statistical models, real-time analysis and algorithms to find patterns in data and deliver instant recommendations. It is present in a number of fleet management applications, particularly those that process a large volume of transactions like fuel.
Electronic payment systems have been used by fleets for more than 30 years to control the locations, gallon and dollar limits of their fuel transactions. The “card controls” do not always stop transactions that fleets deem wasteful or even fraudulent.
Drivers might use company fuel cards to buy premium grade instead of regular gasoline, to purchase at locations with high prices or to occasionally add gallons to their personal vehicles.
Rather than dump fuel transaction data into spreadsheets for analysis to find exceptions, fleets are able to get real-time alerts whenever machine learning algorithms detect patterns in their data that call for immediate actions.
Fueling big data
About one year ago, WEX, one of the largest providers of fuel card and corporate payment systems, formed the WEX ClearView analytics team. With a large volume of fuel transactions, the company saw an opportunity to create new analytical products and services for its customers, says Kurt Thearling, WEX’s vice president of analytics.
WEX is capturing millions of fuel transactions per week, nationwide, from its local and over-the-road fuel card programs. Local fleets use the WEX Fleet card and customers in the OTR market use purchasing systems from subsidiaries EFS and Fleet One.
WEX ClearView has created machine learning algorithms that identify exceptions and present results through an Exceptions Module in the WEX Online portal, he says.
The Exceptions Module has a dashboard-style interface that displays the status of each exception using green, yellow and red indicators.
The exceptions include non-regular fuel purchases, odometer/MPG readings, fuel purchases exceeding vehicle tank capacity and more. Clicking on any of the indicators shows drill-down reports of drivers and transactions that have exceptions.
Algorithms detect fraud events such as when drivers enter incorrect odometer readings at the pump and when purchased gallons do not match the expected distance between fueling events based on a vehicle’s average MPG and tank capacity.
WEX ClearView also uses telematics data in its algorithms that detect fraud. WEX has partnerships with telematics providers to access data, with its customers’ permission. One pattern the algorithms look for is when the GPS locations of service stations do not match the locations of vehicles during fuel transactions, he says.
Including the location of a driver’s home in the real-time analysis is also valuable. Algorithms can detect when a driver takes a company vehicle home and fuels soon after, within a few miles of their home. Patterns in the data might show a fuel transaction occurred while a company vehicle was at the home location and its gas gauge did not change, he says.
This pattern would trigger an alert that a driver is likely fueling a personal vehicle.
“You start to come up with patterns that make sense, but patterns can change over time,” he says.
WEX ClearView is also using real-time analysis to show if drivers are making good purchase decisions. Fleets can view reports that show where drivers purchased fuel with a cost comparison to nearby locations, he says.
The analytics team is working on automated messaging system for fleets to send text reminders to drivers when company policies are violated, like buying premium versus regular gasoline or stopping at higher priced locations.
Thearling plans to include a feature in the messaging system that enables fleets to escalate the alerts. On the first infraction, a text would politely remind the driver of the error. On the second infraction, a message would use all CAPS in the heading; and on the third time the driver would get a text in all CAPS.
To create the machine learning algorithms, WEX ClearView uses business intelligence platforms like Tableau to identify patterns in data, he says.
“Modern data analysis tools for machine learning create an ability to look at data in a more intuitive, interactive way through visualization,” he explains. “You can start to distill the data to see what patters are most interesting.”
For the more complex analysis, WEX ClearView uses statistical programming packages like R and Python. For larger projects that involve huge volumes of data, the team moves the processing to servers in the cloud from Amazon Web Services.