Advanced Random Regret Minimization Models

Slots for supervision of MSc students on general topics, such as stated choice and discrete choice modelling, usually fill up quickly. This means that I will only be able to supervise a limited number of students on these topics.

 

However, additionally I will be able to accept a few highly motivated students who want to work with me on bridging the gap between discrete choice modelling and Machine Learning for choice behaviour analysis. 

 

The ideal student for these projects has:

• Knowledge on Choice behaviour modelling (e.g. followed SEN1221 and or SEN1721)

• Knowledge on Machine learning methods (e.g. minor in Computer Sciences)

• Relevant programming skills (e.g. Matlab, Python, R)

​

There are many possible projects, including:

• Bridging two subfields: Stated Choice Experimental Design and Active Learning

• Complementing Discrete Choice models with Articial Neural Networks

• Using Reinforcement Learning for understanding route choice behaviour

​

If you are interested in doing an MSc project on these topics, please send a brief motivation letter by email, clarifying:

• What topic you are interested

• Intended starting date

• Your relevant experiences (projects, courses, etc.)

• Programming skills (languages)

​

​

************************

​

​

Currently I have two M.Sc. projects available on Mobility-as-a-Service, to be conducted @SkedGo, Sydney.

​

1. Project title: Personalising Mobility-as-a-Service app user-experiences

 
SkedGo is a Sydney-based Mobility-as-a-Service start-up. Their app, called TripGo, (check their free app in the GooglePlay/Apple store) integrates a variety of mobility services (e.g. shared bikes, Uber taxis and own vehicles). The app has over 100,000 frequent users worldwide. The order in which the travel options are presented to the user is based on user-defined settings, which are elicited the first time the app is used by means of a set of sliders. However, the connection between these user-settings on the one hand and their actual preferences and travel behaviour on the other hand is unclear. As a result, it is currently not clear to what extent the order in which the travel options are presented aligns well with the user’s preference.
 
This research aims to personalise the TripGo app user-experience, by improving the ordering of the travel options in the TripGo app.
 
This project can be conducted @ SkedGo, Sydney, or remotely from Delft. It builds forth on work done by a previous M.Sc. student, see this link.

​

​

2. Project title: Who is using my App? Investigating Mobility-as-a-Service app users and their travel behaviour

 
SkedGo is a Sydney-based Mobility-as-a-Service start-up. Their app, called TripGo, (check their free app in the GooglePlay/Apple store) integrates a variety of mobility services (e.g. shared bikes and Uber taxis). The app has over 100,000 frequent users worldwide. While the app records the options that are shown to each user, it is unclear whether the users actually adhere to the shown travel options. This lack of knowledge hampers to further improve the travel information and the user-interface.

This research aims to deepen the understanding of how the TripGo app is used by its users. Specifically, the smartphone's GPS and inertial sensors could be used to determine whether or not the travel options are followed up, or not. However, for privacy reasons the GPS trace cannot the transferred to the data servers. Therefore, this task needs to be done locally – that is, on the smartphone itself.

 This project can be conducted @ SkedGo, Sydney, or remotely from Delft. It builds forth on work done by a previous M.Sc. student, see this link.

 

This project involves considerable software development. Therefore, the ideal student has software development skills, e.g. a computer science background, and ideally also interested in Machine Learning.