An aggregated predictive model for reliable transportation networks based on machine learning

This study explores the application of Machine Learning (ML) algorithms in enhancing the efficiency of intermodal transportation systems. It introduces a new Aggregated Arrival Time Prediction (AATP) model with stacking, which combines different prediction models to consider how different transportation legs (transition of one journey to the next) interact with each other. The research builds up on a thorough analysis of a specific rail route in Central Germany. The stacking model adeptly integrates predictions from diverse transport legs, providing a holistic view of the system and enabling accurate forecasting of arrival times. This aggregation approach is crucial for capturing the complex interdependencies and dynamics within intermodal transportation networks. Through comprehensive data analysis, preprocessing, and the development of base and stacking models, the study effectively enhances transportation performance. The findings underscore the model’s high accuracy in arrival time predictions, showcasing the transformative potential of ML in streamlining efficiency and reliability in modern transportation logistics.

Elham Ahmadi is a PhD candidate and Research Associate at Kühne Logistics University, focusing on Data Science and Machine Learning in Logistics. She joined in May 2022 contributing to the "CargoSurfer" project under supervision of Prof. Dr. André Ludwig and Henrik Leopold. Elham holds two MSc degrees, the latest in Supply Chain Engineering and Management from Jacobs University Bremen, and the other one in Industrial Engineering from Yazd University, Iran. Her expertise includes Forecasting and AI applications, with professional experience as a Quality Assurance Manager in Iran's steel pipe manufacturing industry.