JQJin Qi

Jin Qi

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Jin Qi has a total of 30 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Passenger waiting behavior with sunk waiting time: An empirical study in the ride-sourcing market

    by Yuhan Liu, Jin Qi, Hai Yang, Yili Tang

    Understanding passengers’ waiting behavior is crucial in the ride-sourcing market, as it offers valuable insights into customer preferences, tolerance thresholds, and decision-making processes. During the waiting period, passengers receive updated delay announcements; however, the remaining wait time in these updates may not decrease as anticipated due to the platform’s limited predictive accuracy or unexpected traffic conditions. Such expectation disconfirmation can lead to increased frustration and, in some cases, result in customers abandoning the service. In this study, we perform an empirical analysis of passengers’ waiting behavior in four metropolitan areas in China using a survival analysis model. Our findings reveal an asymmetric response to expectation disconfirmation during the waiting process and display dynamic sensitivity to disconfirmation based on sunk waiting time. Furthermore, we introduce the concept of willingness-to-wait (WTW) to examine passengers’ waiting behavior with the combined effects of sunk waiting time and expectation disconfirmation. The results indicate that WTW follows a non-linear decline and exhibits distinct patterns across different areas and time periods. These insights provide actionable guidance for platforms to optimize subsequent delay announcements, enhancing their ability to improve service quality.

  • Open Access

    Surgery Scheduling under Downstream ICU Risks: A Feature-Driven Robust Optimization Approach

    by Yu Wang, Jin Qi, Jiafu Tang, Yu Zhang

    We study a surgery scheduling problem under limited downstream ICU capacity, where the scheduling decisions are made before postoperative ICU demand is realized, while ICU admissions adapt to patient-specific feature information revealed after surgery. This temporal mismatch creates a misalignment between upstream surgical decisions and downstream ICU capacity, increasing congestion risk and operational inefficiency. To address this challenge, we develop a feature-driven robust optimization framework that integrates predictive modeling with operational decision-making. Our proposed model not only improves surgery-duration prediction but also preserves tractability and interpretability for optimization. Furthermore, to account for downstream congestion risk, we construct a feature-driven representation of ICU demand and formulate a distributionally robust capacity constraint. We show that this constraint can be equivalently transformed into a tractable Conditional Value-at-Risk measure, providing a risk-averse safeguard that effectively limits ICU congestion arising from adverse realizations of length-of-stay uncertainty and patient mix. Numerical studies using a comprehensive data set from a cardiology department in China demonstrate that our approach consistently outperforms traditional predict-then-optimize frameworks. In particular, our framework substantially reduces ICU bed shortages during peak-demand periods while improving operational robustness and enhancing the coordination between surgery scheduling and ICU capacity management.

  • Distributionally Robust Group Testing with Correlation Information

    by Daniel Zhuoyu Long, Jin Qi, Yu Sun, Aiqi Zhang

    Motivated by the need for more efficient and reliable methods of group testing during widespread infectious outbreaks, this paper introduces a novel operational improvement to the widely used Dorfman’s group testing procedure, where a single test is conducted on the pooled sample, followed by individual testing of positive pools. Our method minimizes a weighted sum of testing volume and misclassifications, taking prevalence rates and interindividual Pearson correlation coefficients as inputs, and employs a distributionally robust optimization (DRO) framework to address the ambiguity in the joint infection distribution induced by these correlations. We study two correlation structures within a population. In single-cluster cases, where all subject pairs share equal correlation, we connect our analysis to Nash equilibrium principles and show that higher correlation favors larger testing groups, whereas higher prevalence often calls for individual testing. In multicluster cases, where the population consists of several intracorrelated but interindependent clusters, we highlight the effectiveness of mixed-cluster testing strategies, particularly under low prevalence and correlation. This is a notable addition to the prevailing view that advocates pooling correlated individuals. We provide polynomial-time solutions for both correlation structures and demonstrate the trade-offs and benefits of our DRO approach through a thorough comparison with stochastic alternatives. Using a case study based on a real-world COVID-19 data set, we show that our proposed pooling strategy can save up to 0.1 tests per individual compared with an independence-based pooling scheme and up to 0.5 tests per individual compared with a heuristic pooling strategy implemented in the studied region. This paper was accepted by Jeannette Song, operations management. Funding: This research was supported by the National Natural Science Foundation of China [Grants 72293582 and 72422005], the Hong Kong Research Grants Council [Collaborative Research Fund C6103-20GF, General Research Fund 14210523, 16209923, and 16213424], and the Natural Sciences and Engineering Research Council of Canada [Grants RGPIN-2024-05910 and DGECR-2024-00094]. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2023.04308 .

  • Analyzing Network Formation Using Big Data Analytics: A Study of Degree Distribution, Clustering Coefficient, Diameter, and Assortativity

    by Hao Li, Tianzhuang Xu, Jin Qi, Ying-Ju Chen

    This research presents an innovative approach to network development which goes beyond conventional models that use connection rules for individual agents. We show that agents can produce diverse network structures through uniform edge searches which involve random connection attempts without preference. This model offers a new viewpoint on network formation by emphasizing edges in contrast to nodes. Within this framework, edges are considerably more crucial than those in other network formation models as they represent tangible interactions such as question-and-answer exchanges, commercial transactions, and coauthorships, rather than abstract relationships. We prove that the model converges to a stationary equilibrium degree distribution, implying that in the long term, the probability of observing certain network characteristics stabilizes and becomes time-invariant. Furthermore, we derive closed‐form asymptotic expressions for key network statistics and validate them through both simulations and real‐world data sets. Our research shows that three core dimensions are inherently in competition with one another: (1) quantity (mean degree), reflecting agents’ overall connectivity; (2) quality (clustering coefficient), capturing local cohesion; and (3) equity (skewness of the in-degree distribution), indicating how strongly connections concentrate on a few nodes. Our results show that raising the search times boosts quantity but lowers quality by diluting clustering, whereas raising the secondary connection probability enhances clustering yet decreases equity by concentrating in-degree and intensifying the Matthew effect. These trade-offs reveal essential limits of network formation while providing a flexible framework which connects theoretical insight to practical decisions in network design and management. This paper was accepted by J. George Shanthikumar, data science. Funding: The research of J. Qi is funded by the National Natural Science Foundation of China [Grant 72422005] and the Hong Kong Research Grant Council [Grants GRF 16209923, 16213424, and TRS T32-615/24-R]. Y.-J. Chen acknowledges financial support from the Hong Kong Research Grant Council [Grants C6020-21GF, 16212821, and 16204521]. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2023.00475 .

  • Open Access

    Assortment Optimization with Customer-Centric Fairness

    by Yihua He, Qian Liu, Jin Qi

    The rapid growth of e-commerce platforms has enabled businesses to implement customized assortment strategies, tailoring their product offerings to different customer groups. While these strategies can enhance profit, they often raise concerns about fairness, especially when particular customer groups receive fewer options. Such practices can provoke customer backlash and damage a company's reputation. Motivated by these issues, we propose a fairness-aware framework for assortment optimization that balances revenue maximization with limits on disparities in assortment sizes across customer groups. By introducing cardinality-based fairness constraints, our framework limits assortment-size differences across customer groups through heterogeneous and asymmetric fairness tolerances. Under this framework, we explore two policies: deterministic and randomized assortment. For any number of customer types, we develop exact polynomial-time algorithms for both settings. For the special case of two customer types, we derive sharper structural characterizations and exploit them to design more efficient algorithms. We also establish revenue guarantees under fairness constraints and characterize the value of randomization. In two-group markets, our analysis shows that these constraints may compel sellers to offer more products for disadvantaged groups, reduce the number of products for privileged groups, or apply both adjustments. Our numerical results further show a limited impact on sellers' revenue in the instances considered, making them less likely to resist such regulations. However, policymakers must weigh the trade-offs carefully, as fairness constraints can reduce the overall customer surplus in some cases.

  • Robust Pricing With Asymmetric Distributional Information in Valuation

    by Hongqiao Chen, Yihua He, Jin Qi, Lianmin Zhang

    We study a distributionally robust pricing model aimed at maximizing the worst-case profit under limited knowledge of consumers’ valuation distribution. In addition to mean and variance, we incorporate asymmetric distributional information through semivariance, and then explore the robust optimal posted and randomized pricing strategy. We obtain the robust posted price and show that several series of asymptotic three-point distributions can approach the worst case gradually. The inclusion of asymmetry enables sellers to design three distinct pricing strategies, each corresponding to a different level of asymmetry with explicitly defined thresholds. For the randomized pricing problem, we derive a near-optimal strategy by identifying three key pricing candidates. To avoid the complexity of implementing the optimal randomized price, we also explore a compromise strategy that employs a finite set of pricing candidates, which we formulate as a second-order cone program. Furthermore, we provide numerical evidence highlighting the advantages of incorporating asymmetric distributional information over the mean–variance approach, from both the worst-case profit and actual profit perspectives. Our numerical experiments also reveal that posted pricing is more effective for low-cost products, while randomized pricing is preferable for higher-cost items.


  • Elective Surgery Sequencing and Scheduling Under Uncertainty

    by Xiaojin Fu, Jin Qi, Chen Yang, Han Ye

    Problem definition: We consider a surgery sequencing and scheduling problem with uncertain durations of surgeries in the context of an operating theater. From real data collected from a hospital, we observe the common practice, namely, “to follow,” in which surgeries are conducted sequentially and immediately one after another, according to a specific schedule. Methodology/results: Based on this practice, we propose a mathematical framework to balance the risk of delay and idling using the punctuality index, which takes into consideration both the probability and intensity of delay and idle time. We develop a computationally efficient procedure based on Benders decomposition to derive exact solutions for the problem. The scheduling problem is solvable in polynomial time when the sequence is given. The framework can also accommodate a robust setting when the underlying probability distribution is not fully available. For practical use, we propose two effective heuristics for sequencing decisions by approximating the model. Using real data, we demonstrate that our framework is significantly better than the risk-neutral and probability-maximizing approaches in both performance and computational efficiency. Moreover, the robust setting can effectively lessen the risk of extremely long delay and idle time, and the heuristics are efficient with only a small sacrifice of performance. Managerial implications: With the to-follow policy, our sequencing and scheduling model describes the actual practice better. The two heuristics can be applied easily and directly to help managers of operating theaters to make decisions. Funding: J. Qi is supported by the Hong Kong Research Grants Council [Grants 26202016, 16212419, and 16204521]. Supplemental Material: The online appendix is available at https://doi.org/10.1287/msom.2022.0029 .


  • Top Cited Papers

    Supermodularity in Two-Stage Distributionally Robust Optimization

    by Daniel Zhuoyu Long, Jin Qi, Aiqi Zhang

    In this paper, we solve a class of two-stage distributionally robust optimization problems that have the property of supermodularity. We exploit the explicit worst case expectation of supermodular functions and derive the worst case distribution for the robust counterpart. This enables us to develop an efficient method to obtain an exact optimal solution to these two-stage problems. Further, we provide a necessary and sufficient condition for checking whether any given two-stage optimization problem has the supermodularity property. We also investigate the optimality of the segregated affine decision rules when problems have the property of supermodularity. We apply this framework to several classic problems, including the multi-item newsvendor problem, the facility location problem, the lot-sizing problem on a network, the appointment-scheduling problem, and the assemble-to-order problem. Whereas these problems are typically computationally challenging, they can be solved efficiently under our assumptions. Finally, numerical examples are conducted to illustrate the effectiveness of our approach. This paper was accepted by Chung Piaw Teo, optimization. Funding: This work was supported by the National Natural Science Foundation of China [Grant 71971187], and Hong Kong Research Grants Council [Collaborative Research Fund (C6103-20GF), General Research Fund (14210821, 16204521, 16212419)]. Supplemental Material: The online appendix and data are available at https://doi.org/10.1287/mnsc.2023.4748 .


  • Open Access

    Fragility Index: A New Approach for Binary Classification

    by Chen Yang, Ziqiang Zhang, Bo Cao, Zheng Cui, Bin Hu, Tong Li, Daniel Zhuoyu Long, Jin Qi, Feng Wang, Ruohan Zhan

    In binary classification problems, many performance metrics evaluate the probability that some error exceeds a threshold. Nevertheless, they focus more on the probability and fail to capture the magnitude of the error, which evaluates how large this error exceeds the threshold. Capturing the magnitude of error is desired in many applications. For example, in detecting disease and predicting credit default, the magnitude of error illustrates the confidence in making the wrong prediction. We propose a novel metric, the Fragility Index (FI), to evaluate the performance of binary classifiers by capturing the magnitude of the error. FI alleviates the risk of misclassification by penalizing the large error greatly, which is seldom considered by standard metrics. Moreover, to strengthen the generalization ability and handle unseen samples, we adopt the framework of distributionally robust optimization and robust satisficing, which allows us to derive and control the maximum degree of fragility of the classifier when the distribution of samples shifts. We show that FI can be easily calculated and optimized for common probabilistic distance measures. Experiments with real datasets demonstrate the new insights brought by FI and the advantages of classifiers selected under FI, which always improve the robustness and reduce the risk of large errors as compared to classifiers selected by alternative metrics.


  • Open Access

    Robust Continuous-Time Service Network Design under Travel Time Uncertainty

    by Shengnan Shu, Zhou Xu, Jin Qi

    The continuous-time service network design problem (CTSNDP) has wide applications in the field of transportation, but it is complicated by travel time uncertainty resulting from unpredictable traffic conditions. Incorporating uncertain travel times poses a significant challenge, as time-indexed mixed-integer linear programming (MILP) formulations commonly used to solve the CTSNDP with deterministic travel times become impractical. This is due to their inability to distinguish between decisions that rely on travel times and those that do not. To tackle this challenge, we study a robust CTSNDP under travel time uncertainty, aiming to design a transportation service network with reliable operational efficiency even in the presence of travel time deviations. To incorporate the travel time uncertainty in a tractable manner, we propose a novel consolidation-indexed MILP formulation for the deterministic CTSNDP, eliminating the requirement for time indices. This enables us to derive MILP formulations for both a robust optimization model and a robust satisficing model of the CTSNDP under travel time uncertainty. To solve these formulations exactly, we have developed two tailored column-and-constraint generation methods. Our computational results demonstrate the effectiveness of these solution methods and the tractability of the proposed formulations. Furthermore, the robustness of the solutions obtained has also been verified, and the trade-off between the robustness and its price has been highlighted.