Data Science Ph.D. Dissertation Defense | Data Science | Geri Dimas

Friday, April 14, 2023
1:30 pm to 2:30 pm

United States

Floor/Room #
105

 

DATA SCIENCE 

Ph.D. Dissertation Defense 

Geri Louise Dimas 

Friday, April 14, 2023, 1:30pm-2:30pm

Location: Salisbury Labs 105 

Zoom Link: https://wpi.zoom.us/j/3101917542

Committee members: 

Prof. Andrew Trapp, WPI, Business School / Data Science                                               

Prof. Renata Konrad, WPI, Business School

Prof. Daniel Reichman, WPI, Computer Science                               

 Prof. Kayse Lee Maass, Northeastern, Mechanical and Industrial Engineering  

 

Title: Data Science for Improving Operations in Organizations that Serve Vulnerable Populations

Abstract:  

Data Science for Social Good continues to gain attention in research and the media. Data Science and Analytics can be used in many ways to aid vulnerable sectors of our society. I contribute to this effort by using analytics to improve the operations of governmental agencies, non-governmental organizations (NGOs) and nonprofit organizations serving vulnerable populations in two important domains: anti-human trafficking and immigration. 

In my first investigation, I conducted an in-depth review of the current research landscape of Analytics and Operations Research as applied to the domain of anti-human trafficking. This review provides analysis for understanding, illuminating gaps, and proposing ways forward for those working at the intersection of Applied Analytics and Operations Research to fight human trafficking. In addition, I created an interactive tool that allows researchers to explore further the meta-data associated with my review. In my next investigation, I examined the operational efficiency of an NGO working to fight human trafficking. I provided insights to aid this resource-constrained organization in improving its novel transit-monitoring operations at the Nepal / India border. In my third investigation, I studied the United States growing immigration court backlog by using a queueing-theoretic approach to examine the data, system structures, and behaviors. The derived framework provided an initial understanding of the United States Immigration court system and was used to develop a discrete event simulation model of the New York City immigration court system. Through the discrete event simulation model, I demonstrated how changes in arrival and service rates affect key performance indicators (KPIs) of the system. 

The developed simulation model captures the complexity and interdependencies of the immigration court system and provides a foundation for further evaluation. Motivated by these insights, I extended the discrete event simulation model to enable an in-depth and robust exploration of how policy can impact and improve outcomes. In particular, I incorporated three policies supported by domain experts and evaluated the influence each has on reducing the KPIs of sojourn times, wait times, and queue lengths. The first policy varies the total number of judges and illustrates the impact the quantity of judges has on system throughput. The second policy prioritized asylum cases and evaluated the equity of dedicated dockets. The final policy aimed to minimize sojourn and wait times due to court-caused delays through the introduction of “make-up” capacity. The testing of such policies within my model provided a foundation for data-informed insights for decision makers, something in critically short supply in this important aspect of our society.

 

 

Audience(s)

DEPARTMENT(S):

Data Science
Contact Person
Kelsey C Briggs

PHONE NUMBER: