Intelligent Predictive Models for Crowding on Trains
University of Kent
With support from Transport for London and Rail Delivery Group
The project developed accurate and practical prediction models for rail crowding using an intelligent data-driven modelling method. These models can be merged into the existing software tools to provide customers with detailed crowd prediction information before they and support better decision-making.
The research demonstrated how techniques such as non-parametric forecasting, fuzzy rule based systems and data classification and clustering, can be used to accurately predict daily crowding patterns
Crowding has a major impact on the customer experience which is one of the rail industry 's critical challenges. This research proposed a modelling framework to predict the crowding levels on trains, which will help passengers to plan their journey strategically.