COMP4801 2020-21 Final Year Project

Train delay analysis driven by machine learning

Joint project between MTR and HKU

Introduction

Improve the one of the most busiest metro network

Hong Kong has one of the busiest metro networks in the world, serving over 5 million passengers every day. This extensive metro is also the backbone of Hong Kong transportation that delays and accidents may halt many million citizen activities. MTR company is looking to utilize its high-volume historical data to create an intelligent model to provide a resilient and reliable service for Hong Kong citizen.

Analyse high volume metro data using machine learning

We aim to develop a tool driven by machine learning, to predict the processing time of delays taking place along the Kwun Tong line. This prediction model will help MTR make a prompt and appropriate response to incidents and delays.

Team

Supervisor

Professor Reynold Cheung

Student

Olivier Kwok

Student

Simon Wong

Student

Kalvin Leung

Student

Kayne Chung

PhD Candidate

Wenya Sun

Research Assistant

Nancy Lam

Schedule

Phase 1
Project Conception and Data Exploration
Aug 2020 - Sep 2020
  • Understand previous work
  • Conduct literature review
  • Prepare project plan
  • Create project website
Phase 2
Data analysis and machine learning model design
Oct 2020 - Jan 2021
  • Prepare training dataset
  • Prototype model design
  • Train model
  • Prepare interim report
Phase 3
Interpretation of Results and Refinement of Performance
Feb 2021 - Apr 2021
  • Fine-tune neural network model
  • Collect result data
  • Prepare documations
  • Prepare final report

Documentations

Project Plan

Interim report

Final report