From Data to Predictable Port Operations at ECT Rotterdam
Data Science & AI
The challenge
Developing a predictive model is one thing, but ensuring that the model is actually reliable, manageable, and usable in day-to-day operations requires more.
ECT already had several predictive models in place, including models for the expected container unloading time and truck turnaround time at the terminal. The challenge lay in further improving these solutions and professionalizing the way Data Science solutions were developed, monitored, and deployed into production.
This required constantly bridging the gap between the technical capabilities of data and AI and the day-to-day operations at the terminal.
ECT already had several predictive models in place, including models for the expected container unloading time and truck turnaround time at the terminal. The challenge lay in further improving these solutions and professionalizing the way Data Science solutions were developed, monitored, and deployed into production.
This required constantly bridging the gap between the technical capabilities of data and AI and the day-to-day operations at the terminal.
Our approach
IBS worked as part of a multidisciplinary Agile team on the full lifecycle of various data and AI solutions. In doing so, we combined Data Science, Data Engineering, and MLOps with close collaboration with the operations team.
We analyzed and improved existing predictive models, explored new data sources, and developed new methodologies to better align predictions with operational reality. Among other things, we worked on predicting truck turnaround times and container movements.
At the same time, we streamlined the technical infrastructure surrounding these models. This included improving Python and SQL code, setting up CI/CD processes, and implementing monitoring and alerting within Azure. As a result, not only did the model itself improve, but so did the way models are developed, managed, and monitored.
We analyzed and improved existing predictive models, explored new data sources, and developed new methodologies to better align predictions with operational reality. Among other things, we worked on predicting truck turnaround times and container movements.
At the same time, we streamlined the technical infrastructure surrounding these models. This included improving Python and SQL code, setting up CI/CD processes, and implementing monitoring and alerting within Azure. As a result, not only did the model itself improve, but so did the way models are developed, managed, and monitored.
context
Every day, ECT Rotterdam handles large numbers of containers, trucks, and ship movements at its terminals in the Port of Rotterdam. In this complex operational environment, data is playing an increasingly important role. Reliable forecasts help both terminal employees and customers to plan ahead, better organize processes, and respond more quickly to changes.
Within the MyTerminal customer platform, data and AI solutions are used to make this operational information accessible to end users. IBS supported ECT in the further development and professionalization of this data science environment.
Within the MyTerminal customer platform, data and AI solutions are used to make this operational information accessible to end users. IBS supported ECT in the further development and professionalization of this data science environment.
The impact
The project contributed to a more professional and reliable implementation of data science within MyTerminal. Predictive models could be evaluated and managed more effectively, while monitoring and development standards ensured that issues were identified more quickly.
By combining technical innovation with operational expertise, we developed solutions that not only perform well on paper but, above all, align with day-to-day operations. This gives both employees and customers earlier insight into what will happen at the terminal, allowing them to better adjust their schedules accordingly.
By combining technical innovation with operational expertise, we developed solutions that not only perform well on paper but, above all, align with day-to-day operations. This gives both employees and customers earlier insight into what will happen at the terminal, allowing them to better adjust their schedules accordingly.
Why this case matters
Data and AI only deliver real value when insights actually reach the people who need to work with them. The case at ECT shows that successful AI therefore goes beyond simply developing a good algorithm.
It requires reliable data, a mature technical environment, continuous monitoring, and close collaboration with the business. By bringing these elements together, predictive models can evolve from experiments into an integral part of daily operations.
It requires reliable data, a mature technical environment, continuous monitoring, and close collaboration with the business. By bringing these elements together, predictive models can evolve from experiments into an integral part of daily operations.
Isatis business solutions
From a good model to a solution that truly works in everyday practice. IBS helps organizations develop and refine data and AI solutions and successfully bring them into production.
Curious about what data and AI can do for your organization? Contact us.
Curious about what data and AI can do for your organization? Contact us.
