Are you ready to get into artificial intelligence?Check your basics first

John H. Williams, executive director of corporate data and senior analytics at RaceTrac

RaceTrac’s journey into the world of artificial intelligence and machine learning took about three years to complete. Before we venture into this space, we must assess the current environment to ensure we have a solid foundation. To do this we must evaluate the following:

• Technology stack

• people

• process

From a technology perspective, we moved from a 100% on-premises environment to a hybrid cloud environment. One of our key considerations when planning this move was our employees. We identify our internal technical strengths and weaknesses and provide appropriate training where necessary. We also brought in technology partners to build a strong team.

This assessment involves not just within IT, but also our overall data culture. We have established data and AI education programs to educate all data users on proper data definitions and usage, data visualization, and a basic understanding of our new technology stack (cloud, data lakes, etc.).

“The implementation of artificial intelligence/machine learning is critical to the success of any company. If used correctly, it can increase revenue, reduce expenses, and improve productivity, which may lead to increases in market share.”

Data processing is critical to the success of artificial intelligence projects. Without reliable and trustworthy data, your efforts will be in vain. Therefore, we have established processes and procedures to improve data speed, accuracy and governance (data catalog, data owners and administrators).

After establishing a “solid foundation”, we are confident that we are ready to enter the field of artificial intelligence/machine learning. One of our most successful AI/ML efforts is our Fuel Pumps Down program. This is one of our first AI/ML projects. After completing the deployment of IoT devices on our pumps, we began transferring data instantly to our cloud environment. We use machine learning to identify patterns in each pump handle over a specific time frame. On this basis, we establish control limits based on several variables and factors. Once the fuel pump exceeds control limits, the system establishes a work order for technicians to investigate. This use case started out small in scope, but due to its success it has grown to add additional use cases to address maintenance issues. We have a fleet of approximately 18,000 fuel pumps and until then, 5% of our fleet will be down for maintenance. In less than a year, we got that down to less than 0.1%. This effort has also reduced the number of ticket calls our help desk receives. About 50% of our fuel pump work orders are now created through AI/ML. More importantly, this reduces the workload for our team members, allowing them to focus on creating a better experience for our customers.

Like all initial AI/ML efforts, we encountered challenges and expected them. In addition to new technologies, one of the challenges we face is education. Few within the organization were familiar with the new initiative, raising suspicions. Education and communication are critical to the success of this effort. This education must include all stakeholders, consumers, and project participants. We ensure everyone is properly educated and therefore confident in the success of this endeavor. We also face the challenge of a large number of false positives. False positives are bad, this is an opportunity to improve. Through education and communication, we ensure that all business users understand that AI/ML is not an exact science and that models need to be continuously trained and retrained as business factors change. Some of these false positives revealed business processes that the core team was unaware of, improving business processes as well as data lineage and governance.

We are now adding capabilities such as remotely restarting the fuel pump under certain conditions and evaluating edge computer vision artificial intelligence. The implementation of artificial intelligence/machine learning is critical to the success of any company. If used correctly, it can increase revenue, reduce expenses, and increase productivity, which may lead to an increase in market share. These are just some of the benefits of artificial intelligence/machine learning. However, before you can make this leap, you must have a solid framework and foundation of information.