A Data-Driven Approach to Bus Timetable Optimisation Recommendations

Undergraduate Dissertation, Jonathan Foot

Introduction

In my third year at the University of Nottingham, School of Computer Science I completed my undergraduate dissertation focusing upon optimising bus timetables using a data-driven approach of historical bus timetable open data. The search optimisation algorithm was implemented using Tabu-Search coupled with Squeaky Wheel Optimisation for a more targeted approach to the search space. I attained 85%, a First class dissertation.  The following page outlines all of the work and achievements performed during the project.

Abstract


In this dissertation, I explored using several months worth of historical bus journey data to feed a search algorithm on three main optimisation criteria, with the goal that the program could be used by a bus operator to aid with improving their timetables during a review period. More data than ever before is being recorded about buses within the UK and this provided an exciting opportunity to encourage greater usage of buses, for the benefits they provide.

The first optimisation criteria was minimising unneeded slack time and travel times while balancing the percentage of buses predicted to be on time. The second optimisation criteria was to maximise cohesion between services that share a common route segment, by ensuring they are more spread out at shared stops. A shared route segment is defined as "N" consecutive stops shared by two or more services, where "N" is the minimum segment length. The third optimisation target is to minimise change as much as possible, as too many changes at once make it very difficult to predict how it is likely to perform in the real world. The search algorithm used is tabu-search coupled with squeaky wheel optimisation for a more targeted and informed search. I have demonstrated that tabu-search and squeaky wheel optimisation can be used to optimise a buses timetable to great effect.


Supplementary Material

In addition to the above dissertation I was required to produce the following additional material for submission with my dissertation. These can be found linked below and on the GitHub Repo.

Project Proposal

The original initial proposal for the project which outlines the main aims and objectives and reasonings.

Project Code

Links to the C# source code for the project. Hosted currently on Github.

Demo Video

A 3min long video that summarises the three optimisation criteria and demos the program operating.

Presentation

A 10min long presentation that summarises the project and explains how the search algorithm works.

Demo Video

For the project I was tasked with creating a 3min long video which briefly explains the optimisation criteria of the project and demonstrates the program running and operational.

Presentation

After completing my dissertation I had to perform a 10min presentation and 5min question and answer session about my work. Below shows that presentation, the animations/transitions only partially work on the online version of PowerPoint. Some images have been changed due to copyright reasons.

 Jonathan Foot  © Copyright 2021 - All Rights Reserved

Cookie Policy   • Terms & Conditions  •Contact Me