Abstract
Heritage Community Transportation (HCT) provides essential transportation service in low-income neighborhoods in east Pittsburgh. HCT’s future is in flux due to a 50% drop in ridership since COVID and uncertainty in funding. By the end of an existing project with HCT in June 2022, our team will have proposed a service change for one of HCT’s routes. We plan to collaborate with HCT for another year (July 2022 – June 2023) to fully roll out the deployment in a controlled way, which will help evaluate technical solutions on the ground and adjust quickly to ensure the overall success and continuity of service change. This project also provides opportunities for faculty and students to learn and conduct research in an under-studied area.
Description
Heritage Community Transportation (HCT) provides essential first- and last-mile transportation service in low-income neighborhoods in east Pittsburgh for residents going to / from essential activities such as healthcare appointments, work, grocery shopping, and connection to Port Authority buses. Despite being a vital service provider, the future of HCT is in flux due to a 50% drop in ridership during COVID and uncertainty in future funding.
By the end of an ongoing project with HCT in June 2022, our research team will have finished a thorough data analysis, the design and evaluation of optimization models that compare multiple transportation solutions, operational recommendations, and a plan to run a pilot on one route.
To see through the deployment, we propose to collaborate with HCT for another year from July 2022 to June 2023, in order to fully roll out the deployment in a controlled way across all of HCT’s routes. We propose to have this extra year for deployment since it is not uncommon to see solutions being modified or completely revamped during the deployment phase, especially when circumstances are changing quickly during and post COVID. The deployment-focused phase will help maximize HCT’s service continuity, evaluate our technical solutions on the ground, analyze feedback from riders and drivers, so we can make adjustments and refinements quickly and effectively to ensure the overall success of such service change. This in turn can help improve HCT’s long term financial and operational health. On the academic side, this project will provide ample opportunities for faculty and students to learn in the classroom, share findings at conferences and seminars, and produce high quality academic publications.
Additional details of the project are as follows, outlined in the five-step M’DEAR framework.
Metrics Design:
Before implementing any service change, the research team and HCT will converge on a set of performance metrics, and quantify such metrics for past performance before and during COVID.
In particular, we believe that in addition to the traditional general transportation metrics such as cost per passenger, we need to add additional metrics that measure the criticality of HCT service. For example, how HCT provides essential service to people that are going to / from health care appointments, and whether they have alternative transportation means if HCT were to reduce service level in their neighborhoods. Academically, the metrics should include or inform parameters to calibrate a transportation choice model.
Deployment of Pilot:
Data and performance metric monitoring takes place throughout the pilot deployment of service change on one route. Changes impact schedule, bus stops, communication methods, idle and pass-up choices.
Evaluation of Pilot:
This includes analysis of data and metrics, along with surveys and interviews with the drivers and riders. In particular, we will utilize the Fall 2022 annual HCT survey to get feedback on the pilot route service change. Academically, econometric models can be used to provide causal analysis for the impact of service change. For example, regression discontinuity allows us to use discontinuity in performance metrics time series data to understand the short-term causal impact of service change, while difference-in-difference or synthetic control methods between different routes allows us to discern the medium-term impact of service change.
Adjustment:
By the qualitative and quantitative observations from the pilot deployment, we would be able to obtain and simulate more data on the potential outcomes of different service changes. Academically, this allows us to calibrate a more accurate optimization model.
Re-deployment:
In the final stage of the project, we will then use the tested models and practically verified results to inform final service changes across all HCT routes.
Timeline
To illustrate the distinction between an ongoing project and proposed project, and also show the continuity of the two projects, we provide timelines for both projects below.
Existing project timeline (Month 1 to Month 12)*:
M1-M2 (July, August 2021): Problem definition, literature review, and survey design.
M3-M6 (Sep-Dec 2021): Data collection and data mining with HCT
M3-M6 (Sep-Dec 2021): Deterministic model analysis
M7-M10 (Jan-April 2022): Uncertainty analysis, and robust modeling
M11-M12 (May, June 2022): Propose pilot plan for one route
Proposed project’s timeline (Month 13 to Month 24)*:
M13 (July 2022): Pilot deployment for one route
M14-M18 (Aug-Dec 2022): Monitoring and evaluation of service change. Analysis of controlled rollout of service change
M19-M21 (Jan-Mar 2023): Revised and refined deployment plan
M22-M24 (Apr-June 2023): Full deployment across all three routes
* Monthly or more frequent discussions and feedback with HCT and Delta
Strategic Description / RD&T
Deployment Plan
This project focuses on deployment of operational changes at HCT via a controlled rollout.
We plan to utilize the existing project with HCT to define metrics that are specific and important to HCT’s service. During the new project’s duration, we plan to execute the deployment in four additional steps. We call the five-step plan M’DEAR Deployment Plan.
Metrics Design:
Before implementing any service change, the research team and HCT will converge on a set of performance metrics, and quantify such metrics for past performance before and during COVID.
In particular, we believe that in addition to the traditional general transportation metrics such as cost per passenger, we need to add additional metrics that measure the criticality of HCT service. For example, how HCT provides essential service to people that are going to / from health care appointments, and whether they have alternative transportation means if HCT were to reduce service level in their neighborhoods. More generally, the service change will provide a natural experiment for us to observe the parameters of a transportation choice model.
Deployment of Pilot:
Data and performance metric monitoring the pilot deployment of service change on one route. Changes impact schedule, bus stops, communication methods, idle and pass-up choices.
Evaluation of Pilot:
Analysis of data and metrics, along with surveys and interviews with the drivers and riders. In particular, we will utilize the Fall 2022 annual HCT survey to get feedback on the pilot route. Econometric models can be used to provide causal analysis for the impact of service change. For example, regression discontinuity allows us to use discontinuity in performance metrics time series data to understand the short-term causal impact of service change, while difference-in-difference or synthetic control methods between different routes allows us to discern the medium-term impact of service change on performance metrics.
Adjustment:
By the qualitative and quantitative observations from the pilot deployment, we would be able to obtain and simulate more data on the potential outcomes of different service changes, which allows us to better calibrate an optimization model.
Re-deployment:
In the final stage of the project, we will then use the tested models to inform new service changes across all HCT routes.
Expected Outcomes/Impacts
Paper submissions by the end of the project
• One paper for method-oriented journal (e.g., Operations Research), focusing on solution concepts and equilibrium analysis in a robust game theoretical model between service provider and riders
• One paper for practice-oriented academic journal (e.g., INFORMS Journal on Applied Analytics), focusing on the appropriateness and practicality of transportation solutions in different transportation contexts, especially in low-income neighborhoods (e.g., static vs. flexible/on-demand schedules).
• One paper that extrapolates the findings from this project to a national level analysis, targeting a high-level, broad-audience science and policy journal (e.g., PNAS, Science). This paper focuses on the gaps and opportunities of last-mile transportation as a social determinant of employment opportunities and / or health outcomes in low-income neighborhoods across the country.
Potential conference presentations and seminars
• Smart Mobility Connection seminar
• Traffic21/Mobility21 Deployment Partner Consortium Symposium
• Production and Operations Management Annual Conference (Summer 2022)
• INFORMS Annual Conference (Fall 2022)
Education and Training
• One postdoctoral researcher to lead one or two aforementioned papers
• One PhD student (funded by Heinz) to co-author one or two papers
• One or two masters level research assistants at Heinz College to perform data analysis, and learn about transportation problems and solutions
• One or two undergraduate / masters level research assistants from Civil and Environmental Engineering during the 2022 CEE summer research program
• One teaching assistant to develop a case study for Heinz/CEE course: 94-867/12-768 Decision Analysis for Business and Policy by Spring 2023.
Expected Outputs
TRID
Individuals Involved
| Email |
Name |
Affiliation |
Role |
Position |
| pyzhang@cmu.edu |
Zhang, Peter |
Carnegie Mellon University |
PI |
Faculty - Untenured, Tenure Track |
Budget
Amount of UTC Funds Awarded
$100000.00
Total Project Budget (from all funding sources)
$198350.00
Documents
| Type |
Name |
Uploaded |
| Data Management Plan |
DMP_1e6V2Vc.pdf |
Nov. 18, 2021, 7:01 p.m. |
| Project Brief |
Slides_for_Project_400_-_Controlled_Deployment_of_Analytical_Solutions_for_Essential_Transportation_Services_in_Low-Income_Neighborhoods.pptx |
March 4, 2022, 9:29 a.m. |
| Publication |
Adjustability in robust linear optimization |
Oct. 14, 2022, 12:38 p.m. |
| Publication |
Structure of social welfare functions |
Oct. 14, 2022, 12:38 p.m. |
| Publication |
TSP/VRP Approximation for Assessing Urban Transportation Efficiency |
Oct. 14, 2022, 12:38 p.m. |
| Presentation |
Demand learning and supply optimization for last mile transportation in low-income neighborhood |
Oct. 14, 2022, 12:38 p.m. |
| Presentation |
BHH Theorem Through a Big Data Lens: TSP/VRP Approximation for Assessing Urban Transportation Efficiency |
Oct. 14, 2022, 12:38 p.m. |
| Presentation |
Adjustability in Robust Linear Optimization |
Oct. 14, 2022, 12:38 p.m. |
| Progress Report |
400_Progress_Report_2022-09-30 |
Oct. 14, 2022, 12:38 p.m. |
| Progress Report |
400_Progress_Report_2023-03-30 |
April 7, 2023, 10:02 a.m. |
| Publication |
Fleet Sizing and Allocation for On-demand Last-Mile Transportation Systems |
April 10, 2023, 9:04 p.m. |
| Publication |
Structural Characteristics of Equitable and Efficient Distributions |
April 10, 2023, 9:05 p.m. |
| Publication |
Joint Optimization of School Bus Routes and Last-mile Services |
April 10, 2023, 9:05 p.m. |
| Publication |
Dynamic and Robust Network Resource Allocation |
April 10, 2023, 9:06 p.m. |
| Final Report |
Final_Report_-_Zhang_400.pdf |
Sept. 15, 2023, 10:03 a.m. |
Match Sources
No match sources!
Partners
| Name |
Type |
| Heritage Community Initiative |
Deployment & Equity Partner Deployment & Equity Partner |
