Skip to main content

Transportation Resources For Carriers Like You

The Integral Role of Data Science in Transportation Technology

When we talk about the transportation and logistics industry, we often speak of it in terms of trucks, freight, drivers and miles; all tangible, straightforward elements of the business. But the numbers behind transportation – data – play an ever-increasing role in every decision a company makes. As a result, “data science,” has become integral to transportation companies’ daily operations. 

But what is data science? How does a company take advantage of the insights it affords? And how does Trimble deliver its own data science expertise? We spoke with Trimble’s data science team about the history of data science, its importance for the transportation industry and how Trimble uses data science to design and enhance its solutions.

Understanding Data Science 

When you hear the term “data science” you might assume it’s a never-ending compilation of numbers and percentages only meant to cause business decision makers more stress. But really, it’s a key internal process at the core of every business, and everyone inside and outside the business plays a part.

Every day, you consume and participate in data science. From car insurance companies saving you “15% or more,” 9 out of 10 dentists recommending a toothpaste brand or hand sanitizer killing 99.9% of all germs, data is constantly buzzing by your ears. When you choose to get your lunch from one business over another, you are actively participating in the world of data science. Just by living your life, you’re contributing to the economic losses and gains of businesses. A data scientist’s job is to gather this data, analyze it and predict the trends to follow for the benefit of the business.

Trimble Principal Data Scientist, Clint Vrazel, explains it in simple terms: “Data science to me is a huge toolkit to manage, understand, learn from and leverage data. It includes seeing what is going on, what could happen and what should happen. This could include statistical models, algorithms, machine learning and artificial intelligence.”

Data science is often referred to as a “new industry,” but it’s been a core component of business activities for centuries. Today’s data science includes a list of techniques pulling from economics, genomics, robotics, linguistics, psychology, physics, etc. Virtually all modern companies are data-focused. “Eventually, every company becomes a ‘data company,’just from the volume, rate and variety of data they produce,” says Vrazel. “So there’s opportunity everywhere to use data - external, internal, public, client - to create or improve products and processes.”  

It is a trial and error type of process, taking what is known and using it to predict what isn’t. Vrazel explains that, “what we know is what has and hasn’t succeeded. What is still unknown is whether any such project will succeed, so it’s best that we try many things, fail fast, get feedback and iterate. When it comes to data science at Trimble, it’s up to those working in the industry to be patient, opportunistic and transparent communicators of what we have, are and can do.”

How Trimble Practices Data Science

There are two different ways Trimbe uses data: first, to design and enhance our Trimble solutions, and second, to find solutions to customer problems. Both processes depend on a deep understanding of a transportation company’s core problem. To gain this understanding, we turn to data. 

The transportation industry generates a massive amount of data. Good, bad, up, down; whatever a company has done across its years of operation, the data can reflect it down to the mile driven, dollar spent, deal closed and drop of fuel consumed. Our data scientists collect this data from a variety of outlets, including directly from the companies we work with, or from third-party vendors that specialize in data aggregation. 

Once the data is collected, it’s time to make sense of it and how the data reflects or impacts the problem a customer is facing. “Most of my time is spent in understanding the business problem and data cleaning, a good 60-70%, then most of the remaining is spent coding, organizing, debugging, documenting,” Vrazel says. “But the process is rarely a straight line. At every step, there’s backtracking to better understand a nuance or complexity.”

This entire process can be summed up by the traditional data science methodology, CRISP-DM. “Data scientists have a complex and creative job. We generally refer to the CRISP-DM process to highlight the responsibilities and process of creating a solution for our customers,” says Donnie Wheat, Director of Data Science at Trimble.

Questions at Every Stage

1. Business understanding – What does  the business need?
2. Data understanding – What data do we have / need? Is it clean?
3. Data preparation – How do we organize the data for modeling?
4. Modeling – What modeling techniques should we apply?
5. Evaluation – Which model best meets the business objectives?
6. Deployment – How do stakeholders access the results?

The CRISP-DM Process - Source: Data Science Process Alliance

   

To ensure an accurate and streamlined process, data integrity and cleanliness is key. Trimble analysts/scientists have come up with a few techniques. The main one being enforcing data format contracts, so that data is always received the same way. Still, Trimble’s solutions are used by a wide variety of customers and partners, so over time we’ve developed a strong set of default values that can be used, as well as covering the most common options.

Non-uniform data can be the real challenge, “potentially insurmountable” according to Vrazel. In the transportation industry, dozens or hundreds of customers  may have data points with the same name and location - and no two are exactly the same! “Often we start development with a chicken-and-egg problem - few customers would use a solution that makes errors, but we need more data to improve the solution.”

As our data scientists work through this process, they are working toward a solution for the customer. “For instance, if you want to optimize a driver’s day shift, you have to understand the orders, trucks, drivers and driver restrictions.  This will help guide you to the data to collect. The patterns to identify are a result of gaining access to that data and understanding,” says Wheat.

How Do We Use Data Science to Improve Customer Experience?

In today’s fast-paced world, it’s not easy for customers to pinpoint their exact problems. They may not know what the problem is, or the answers they’re seeking, because the global supply chain in which they operate is complex and always evolving. 

For example, Sr. Data Scientist Snehal Vartak says that one promising product poised to significantly impact our transportation customers is Tender Evaluation. A product that helps carriers book the right freight at the right time, based on their historical data.

“It enhances the tender acceptance process, resulting in swifter and more informed decisions.” she says. Tender Evaluation leverages an adaptive and explainable AI technique known as Fuzzy Logic. “In contrast to traditional ML algorithms, Fuzzy Logic operates on a rule-based approach that can be readily defined by domain experts. This affords a higher level of control and customization in the process, allowing for improved transparency and fine-tuning.”

This is just one example of the many ways in which data science has been used to improve and better understand the customer experience. Data science, while being all about numbers and facts, can also be considered a very personal process.

“Any optimization we can create in transportation ultimately has direct benefits we see every day, and for that I am very proud of the many transportation optimization solutions that Trimble offers,” says Wheat. “When you look at solutions like Expert Fuel and Appian, you can give a driver an optimum route with the least cost of fuel for the truck to do a trip. That one trip results in a reduction of wasted miles, with a reduction of fuel usage and a savings of fuel cost to the company and ultimately the consumer.”



Want to learn more about how Trimble's data science capabilities can help your company optimize its operations? Contact us for a consultation.