INFORMS 2021 Program Book

INFORMS Anaheim 2021

TC10

TC06 CC Room 303A In Person: Analytics for Policing and Urban Public Service Operations General Session Chair: He Wang, Georgia Institute of Technology, Atlanta, GA, 30332- 0205, United States 1 - Data-driven Optimization for Atlanta Police Zone Design Shixiang Zhu, Georgia Institute of Technology, Atlanta, GA, 30318-2990, United States We present a data-driven optimization framework for redesigning police patrol zones in an urban environment. The objectives are to rebalance police workload among geographical areas and to reduce response time to emergency calls. We develop a stochastic model for police emergency response by integrating multiple data sources, including police incidents reports, demographic surveys, and traffic data. Using this stochastic model, we optimize zone redesign plans using mixed- integer linear programming. Our proposed design was implemented by the Atlanta Police Department in March 2019. By analyzing data before and after the zone redesign, we show that the new design has reduced the response time to high priority 911 calls by 5.8\% and the imbalance of police workload among different zones by 43\%. 2 - Optimizing Shift Schedules and Dispatch of Safety Patrol Officers for Denver Public Schools Amanda Chu, ISyE Georgia Institute of Technology, Atlanta, GA, United States, Pinar Keskinocak, Onkar Kulkarni, Ritesh Ojha Each year, the Safety Department at Denver Public Schools (DPS) manually creates patrol officer schedules to respond to calls for over 200 schools to ensure the safety of all students and staff. The Safety Department struggles to adjust schedules in response to changes such as available officers due to the manual process. To address these drawbacks, we developed optimization and simulation models to create officer shift schedules based on call demand, factor in call demand uncertainty, and estimate the performance of the generated schedules. The DPS Safety Department used one of multiple generated schedules for the 2019-2020 academic year and we able to meet their target call response times. TC07 CC Room 303B In Person: Crowdfunding and Platform Economics in Social Media General Session Chair: Zhen Fang, University of Washington, Seattle, WA, 98105, United States 1 - Opinion Leader Identification Associated With Covid-19 in Online Social Networks Behnam Malmir, Virginia Tech, VA, United States Safety culture is a collection of beliefs, attitudes, and practices that is focused on improving individual and organizational health and safety. Providing interactions between citizens and governments could elevate this improvement. Social media has been known as one of the most valuable tools to this aim and government agencies eagerly have applied social media applications to enhance citizen engagement in managing crises. However, employing opinion leaders (OLs) to improve the effects of social media on people’s safety culture has been neglected in the literature. This paper provides a framework for OLs identification in the era of pandemic situations called ‘Pandemic OL Finder’. The proposed framework comprises three steps of finding potential OLs on the Twitter platform, determining the characteristics of those OLs, and discovering real OLs. 2 - More Than Double Your Impact: An Empirical Study of Match Offers on Charitable Crowdfunding Platforms Zhen Fang, University of Washington, Seattle, WA, 98105, United States, Xue Tan, Shengsheng Xiao, Yong Tan To promote charitable giving, charitable crowdfunding platforms adopted match offers, whereby leadership donors match the others’ donations at a given rate. Our study seeks to understand how the suppliers (donors) evaluate projects with and without match offers differently, especially varying with their donation experience, and how the demanders (fundraisers) react to match offers. At an individual level, we find that, on average, donors derive a higher utility toward matched projects. Warm-list donors are three times more likely to contribute to matched projects, while cold-list donors are twice more likely to do so. New donors prefer unmatched projects. The market-level analysis shows that increasing the matched project ratio benefits both sides of the market. Our work connects micro-level and macro-level to disentangle the impact of match offers systematically.

TC09 CC Room 303D In Person: Trending Topics in Applied Probability General Session Chair: Anton Braverman, Northwestern University, Evanston, IL, 60208, United States 1 - Stability, Memory, and Messaging Tradeoffs in Heterogeneous Service Systems Martin Zubeldia, Georgia Institute of Technology, Atlanta, GA, United States, David Gamarnik, John N. Tsitsiklis We consider a heterogeneous distributed service system, consisting of N servers with unknown and possibly different processing rates. Jobs with unit mean arrive as a renewal process of rate proportional to N, and are immediately dispatched to one of several queues associated with the servers. We assume that the dispatching decisions are made by a central dispatcher with the ability to exchange messages with the servers, and endowed with a finite memory used to store information from one decision epoch to the next, about the current state of the queues and about the service rates of the servers. In this setting, we study the fundamental resource requirements (memory bits and message exchange rate) in order for a dispatching policy to be always stable. 2 - Dynamic Pricing and Assortment under an Unknown MNL Demand Noemie Perivier, Columbia University, New York, NY, United States, Vineet Goyal We consider dynamic multi-product pricing and assortment problems under an unknown demand over T periods, where in each period, the seller decides on the price for each product or the assortment of products to offer to a customer who chooses according to an unknown Multinomial Logit Model (MNL). We propose a randomized dynamic pricing policy based on a variant of the Online Newton Step algorithm that achieves a near-optimal regret guarantee under an adversarial arrival model. We also present a new optimistic algorithm for the adversarial MNL contextual bandits problem, which achieves a better dependency than the state- of-the-art algorithms in a problem-dependent constant. 3 - Distributionally Robust Batch Contextual Bandits Nian Si, Stanford University, Stanford, CA, 94305, United States, Fan Zhang, Zhengyuan Zhou, Jose Blanchet Policy learning using historical observational data is an important problem that has found widespread applications. However, existing literature rests on the crucial assumption that the future environment where the learned policy will be deployed is the same as the past environment that has generated the data—an assumption that is often false or too coarse an approximation.In this paper, we lift this assumption and aim to learn a distributionally robust policy with incomplete (bandit) observational data. We propose a novel learning algorithm that is able to learn a robust policy to adversarial perturbations and unknown covariate shifts. We first present a policy evaluation procedure in the ambiguous environment and then give a performance guarantee based on the theory of uniform convergence. TC10 CC Room 304B In Person: Data Driven Decision Making for Agriculture General Session Chair: Saeed Khaki, Iowa State University, Ames, IA, 50010, United States 1 - A Multi-objective, Soft Constraint Solution To A Capacity- constrained Corn Planting Schedule Problem Mingshi Cui, Miami University, Oxford, OH, 45056, United States Mingshi Cui, Rutgers University, New Brunswick, NJ, United States, Kunting Qi Our research describes a general solution to optimize the planting schedule for corn population seeds that attempts to minimize the median and maximum absolute deviation from location storage capacity, as well as minimizing the number of nonzero harvest weeks while respecting planting windows and weekly harvest capacities. We used a Long Short-Term Memory model to predict daily GDUs, based on historical daily GDUs data. The Genetic Algorithm with multi- objective function and soft constraint models has been implemented based on the predicted values. All of the models’ parameters have been tuned to get the optimized corn planting schedule through an innovative tree-based algorithm.

113