- Time: 13:30-14:30
- Location: Institute for Transport Studies, room 1.11, University of Leeds
An artificial neural network based method to uncover the Value-of-Travel-Time distribution.
Tuesday, 19th of March, 13:30 to 14:30.
Institute for Transport Studies, room 1.11, University of Leeds
In this presentation, we propose an Artificial Neural Network (ANN) based method to derive the Value-of-Travel-Time (VTT) distribution. This method is highly flexible and complements recently proposed nonparametric methods to calculate the VTT from choice data. The strength of this method is that it is possible to uncover the VTT distribution (and its moments) without making strong assumptions about the shape of the distribution or the error terms, while being able to incorporate covariates and account for panel effects. Therefore, in contrast to other nonparametric methods, the ANN-based method is suitable to derive estimates for the mean VTTs for appraisal. In this presentation, we first show the results from a series of Monte Carlo experiments. Using these, we assess how well the ANN-based method works in terms of being able to recover the true underlying VTT distribution. After having demonstrated that the method works well on Monte Carlo data, we apply our method to data from the 2009 Norwegian VTT study. Finally, we extensively cross-validate the method by comparing it with a series of state-of-the-art discrete choice models and nonparametric methods. Based on the encouraging results we have obtained, we believe that there is a place for ANN-based methods in future VTT studies. Finally, we discuss explore further directions to capitalise these machine learning methods for VTT research, and more generally for choice behaviour analysis.
Sander van Cranenburgh is Assistant Professor in the Transport and Logistics Group of Delft University of Technology. His research focusses on methods for choice behaviour analysis. During his Postdoc he worked on Random Regret Minimization (RRM) based choice models, and made several methodological contributions to this field, including the development of new regret models and efficient design theory for RRM models (which has recently been implemented in Ngene).His current research mainly focusses developing data-driven modelling approaches for choice behaviour analysis. In this research he specifically seeks the edge between theory-driven approaches, such as discrete choice models, and data-driven approaches, such as Artificial Neural Networks.
Marco Kouwenhoven is Research Leader at Significance, an independent research institute specialised in quantitative research on mobility and transport, and a part time Researcher in the Transport and Logistics Group of Delft University of Technology. One of his research fields is stated preference research, and especially to determine the value of travel time, value of travel reliability and value of comfort. He has worked on many stated preference projects, among which the latest value of time studies in the Netherlands and Norway.