Holger Köhler, who holds a doctorate in economics, heads a team consisting mainly of business consultants who advise internal units and customers on matters relating to data analytics and develop data-driven use cases. Photo: DB Schenker
Mr. Köhler, for the uninitiated, could you explain briefly what predictive analytics is all about?
In simple terms, it is about making predictions for the future. By evaluating large amounts of historic data available within companies as well as from external sources, a mathematical model can be developed that is capable of identifying patterns in the data. The model can then be used to predict future events based on these patterns – so that we can develop solutions resulting in greater efficiency and lower costs, for example. Predictive analytics has its origin in statistics and is used as a method within the field of machine learning, which in turn is part of artificial intelligence.
What kind of data is being evaluated to this end?
To use an example that is pertinent to DB Schenker: in Europe-wide land-based transportation alone, data for around 100 million shipments is generated annually, and this can then be subjected to complex analyses. The relevant external input data ranges from economic data collected by research institutions to ocean freight carriers’ schedules as well as traffic and weather information. Unstructured data obtained from news channels or from the world wide web can also be included.
How is this data used to produce forecasts that add value?
To stay with land-based transport: we are in the process of rolling out a tool for making predictions and optimizing routes at our 400 terminals across Europe. Based on dispatch data, the tool allows us to predict, with up to 95 percent accuracy, the incoming and outgoing freight volumes that will need to be handled in the coming days and weeks. We can then adjust our capacities as well as the planning of collection-and-delivery tours accordingly. Or looking at Supply Chain Risk Management, for example: by enriching consignment tracking & tracing information with data about traffic volumes, the effects of weather conditions, waiting times at border crossings or strikes in ports, we can compute risk forecasts – and proactively take countermeasures.
How many shipments will there be arriving at this terminal next month? Large amounts of historic data help DB Schenker’s analysts to find answers. Photo: Matthias Aletsee
Can you give other examples?
Well, we use predictive analytics and artificial intelligence in ocean freight to predict market and capacity developments. The aim is, among other things, to provide customers with freight capacities at competitive rates during peak periods, and to do so in good time. We’ve also developed an early warning system that allows us to identify customers who may be considering going elsewhere ahead of time, so that we can implement countermeasures. While this is primarily for our own benefit, it helps the customer as well if we take a close look at areas where there is scope for improvement and then adapt more closely to the customer’s needs. To sum it up: the evaluation and linking of data helps us optimize logistics processes, and it also lets us process more information about the market and about customers, and this in turn makes it possible to deal with customers in a more targeted manner.
Who is currently working with predictive analytics at DB Schenker?
We have a team focusing on data strategy and analytics, including predictive analytics. The team’s work is cross-divisional, and the members are business consultants, data scientists, operations research specialists and big data engineers – job profiles that are new to DB Schenker. Our experts, most of whom have a doctorate, are mainly from the fields of mathematics, physics, computer science and engineering.
Will such methodologies ever be able to replace the experience and expertise of logistics specialists – and should they in the first place?
The fact is that even today, algorithms will often produce better solutions, and more quickly, than humans can. It is now possible to make decisions that are based not merely on gut feeling, but backed by hard data as well. Given the complexity and volatility of the markets, that is a definite advantage. However: at this stage predictive analytics is not intended to replace human expertise, but to supplement it. This will allow employees to focus on matters that are more mentally challenging than many of the run-of-the-mill tasks that still take up a lot of time at present. Added to that is the fact that we are still a long way from having a general artificial intelligence that also includes emotional intelligence – something that is needed in dealing directly with customers.
How do you plan to further advance this topic?
Predictive analytics – or more generally, data analytics including artificial intelligence and self-learning algorithms – are core components of DB Schenker’s digitalization strategy. The plan is to develop the company into a data-driven service provider, capable of offering services that go far beyond the traditional forwarding business. In the future, we will therefore use predictive analytics to support decision-making in all areas: for land-based transportation, air and ocean freight as well for contract logistics, and also including overarching functions such as finance and sales.