Case Study: Commuter cycling policy in Singapore: a farecard data analytics based approach

City:	Singapore
Organization:	Singapore-MIT Alliance for Research and Technology (SMART) & National University of Singapore (Department of Industrial and Systems Engineering)
Project Start Date:	April 2011
Project End Date:	January 2014
Reference:	Kumar, A., Nguyen, V.A. & Teo, K.M. Ann Oper Res (2016) 236: 57. https://doi.org/10.1007/s10479-014-1585-7
Problem Description:	Peak-hour week-day traffic congestion in Singapore can be alleviated by promotion of commuter cycling. This paper takes a data analytics approach to propose policies to promote commuter cycling in most needed areas in Singapore. The main problem is to identify areas where short trips to the MRT stations are occurring and enhance the cycling infrastructure.
Technical Solution:	· Develop a binary linear programming optimization-based decision support model to make an efficient choice of policies/projects to maximise potential cyclists for a given investment level, with inputs such as: cycling demand numbers, cost estimates of cycling infrastructure, percentage switch to cycling from different modes for first-mile and end-to-end trips and investment levels. · Through farecard trip analysis, spatial distribution of first-mile and end-to-end trip centered around MRT stations to identify areas where cycling infrastructure will be most beneficial
Datasets used:	· The fare card database, which stores all public transportation trips during the day, between April 11-15 2011 · Cycling demand – 10% is used for the model, based on literature review · Cost of developing cycling towns – Average cost taken from LTA report, which estimates expenditure costs of new cycling infrastructure · Budget available – 3 different values used to provide a range for different sized towns
Outcome:	· Develop cycling infrastructure around 19 of the 120 MRT stations, which account for 71% of the first mile trips, reducing number of feeder busses and cars travelling to these stations. · Origin and Destination (OD pair) analysis showed that 70% of students’ trips are less than 3 km in length, with majority of the pairs connecting MRT stations, identifying hot spots around the city where to locate cycling infrastructure Policy Recommendations from spatial data analysis: · Promotion of cycling towns to emphasize funding for first-mile bike trips due to its impact on traffic congestion. · Planning cycling regions based on end-to-end OD analysis, growing the cycling region. · Concept of school cycling enclaves in areas where a high proportion of end-to-end cyclists are students Decision Support Model: · Optimization model to support the policy makers in making better choice of cycling towns and regions. · Maximizes the cycling demand subject to various constraints such as ensuring demand is satisfied, remaining within the budget constraints, no open-ended paths and capturing existing cycling towns
Issues:	· Lack of data concerning cycling demand, hence relying upon estimation based on historical data. · Simplification of original optimization model due to lack of data and inability to quantify certain benefits · Missing data for trips made by cars, henceforth approximating the modal-share value used throughout the study · Data only captured for 5 week days but does not account for seasonal effects, skewing the results in favour of cycling
Status of Project:	COMPLETED
Created by:	Ronauq Sabharwal (ronauq.sabharwal@mail.utoronto.ca)