Predicting the Unpredictable:: How Data Can Help Cargo

Predicting the weight of incoming and outgoing air cargo is a complex task. Air freighters rarely carry the exact same cargo repeatedly, and shipments are often seasonal. Triangle flight patterns and multi-dimensional capacity add to the challenge. The specific contents of a flight can also drastically change the capacity of a freighter flight. But the ability to predict cargo weight can go a long way to making the air cargo chain function more smoothly. So Schiphol asked Jacco Fijnheer to study the possibilities: can cargo movements be effectively predicted, and thus easier to manage? Jacco found the answer: with the help of machine learning, they can.

Relationships to build on

Jacco performed the study in order to complete his Master’s degree in Econometrics. But he’d become acquainted with Schiphol Cargo long before that. ‘Two years ago, I completed my first Master’s degree in Transport & Supply Chain Management,’ Jacco explains. ‘When I decided to expand my specialisation with an Econometrics Master’s, I needed to complete a pre-Master’s first. That pre-Master’s programme brought me in touch with Bart Pouwels (Head of Cargo) and Roos Bakker (Director Business Development), as we conducted a consultancy project to study the flower supply chain between Colombia and the Netherlands. When it came time for my Master’s thesis project, I called Bart and Roos to talk about the opportunities at Schiphol.’

Tackling the challenges

And since improving Cargo chain efficiency is a top priority at Schiphol, Bart and Roos helped Jacco find an opportunity to address a big challenge: find an accurate, practical and simple way to predict cargo weight on full-freighter flights in and out of Schiphol. ‘I only had three months to complete the study,’ Jacco says. ‘So I knew I couldn’t go into the kind of depth of typical nine-month qualitative study. Still, I felt confident that we could make great steps towards finding an answer to the predictability question.’

And find an answer, he did. In just three months’ time, Jacco used existing algorithms and machine learning to develop the framework for predicting cargo weight. ‘The limited length of the study meant we could only work with a limited number of variables and parameters. And of course, there are many, many variables in the air cargo chain – seasonality, the number of fixed flights, macroeconomic variables like Gross Domestic Product, the coronavirus, and so much more,’ explains Jacco. ‘But the framework I developed is proven to work, and we’ve established a simple way for Schiphol to continue to add variables and parameters, so that they can eventually predict cargo weight.’ The more the framework is used and fed with new, high-quality data, the better it will become at predicting cargo weights.

Reaching the goal

Of course, it’s not likely that air cargo will ever become 100% predictable. There are simply too many variables to take into account. However, Schiphol believes that technology is key to anticipating the needs of the Cargo community, and to creating smoother, more efficient air cargo operations. ‘And that starts within Schiphol itself,’ Jacco says. ‘In addition to these kinds of studies, Schiphol is working hard to improve the quality and management of the data they have available. It’s a challenge every organisation struggles with – it’s not easy to keep data robust and well managed. But as they improve, Schiphol will begin to see the profound benefits.’

The Cargo community also plays an essential role, according to Jacco. ‘Most Cargo companies are very hesitant to share their company data, because they fear it will jeopardise their competitive edge,’ says Jacco. ‘But I am certain that if Cargo companies have the courage to give it a try, they will see that any short-term risks will be more than compensated by the long-term benefits of a more efficient air cargo chain. Schiphol will be able to prove – in black and white – how sharing data results in benefits for individual Cargo companies, and for the entire chain as a whole.’

Taking the first steps

Jacco’s advice to the Cargo community? Get involved soon, and be willing to explore the possibilities. ‘The Cargo companies that get involved in these data projects early will be the first to benefit from the innovations that lead to a smoother cargo process. And the only way for the whole chain to improve is if the whole chain works together to make it happen. So, I encourage the Cargo community to explore the possibilities with Schiphol as soon as they can.’