Abstract
dc:description.abstractThis dissertation investigates how network structures, prevalent in ridesharing marketplaces as well as social and economic domains, challenge and enhance traditional methods in dynamic pricing and causal inference. The research is unified by the theme of addressing network effects to design better decision policies and obtain more precise estimations of treatment effects. The first part of this thesis addresses dynamic pricing problems in ridesharing platforms like Uber, Lyft, and DiDi, which face supply-demand imbalances across interconnected geographic locations. It proposes a computationally efficient approximate linear programming approach that adjusts prices based on localized demand and supply conditions, thereby optimizing platform revenue. The second part focuses on causal inference under network interference, where traditional assumptions of independence in treatment effects fail due to interactions among units. Recognizing the biases introduced by standard A/B tests, this thesis develops a novel mixed randomization design, blending cluster-based and Bernoulli randomizations to effectively capture the full interference effect. This integrated approach yields an unbiased estimator for the total treatment effect (TTE). Lastly, the thesis examines the pseudo-inverse estimator method for estimating TTE in settings characterized by approximate network structures. New theoretical insights, including variance bounds and an associated variance estimator, are established, significantly advancing the theoretical understanding of this estimator. Practical applicability is validated through an empirical analysis involving a large-scale experiment with over 50 million users.
Degree
thesis:*- Level thesis:degree_level
- Doctoral
- Department dc:contributor.department
- Industrial and Systems Engineering
- Grantor dc:publisher
- Georgia Institute of Technology
- Year dc:date.issued
- 2025
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Jiang, Yiming
- Advisor dc:contributor.advisor
-
- Wang, He
- Committee members dc:contributor.committeemember
-
- Pananjady, Ashwin
- Xie, Yao
- Zhao, Jinglong
- Yu, Christina Lee
Subjects
dc:subject × 3Rights
- Language dc:language.iso
- en_US
Identifiers
dc:identifier.*- Handle dc:identifier.uri
- https://hdl.handle.net/1853/77816