Xuecheng Tian
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Xuecheng Tian has a total of 57 co-authors. Here are the top co-authors by number of shared publications:
- Shuaian Wang
Faculty of Business, The Hong Kong Polytechnic University, Hung Hom 999077, Hong Kong
18 shared publications Shuaian Wang
Faculty of Business, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong 999077, China
14 shared publications- Yong Jin
Faculty of Business, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong 999077, China
7 shared publications - Bo Jiang
Faculty of Business, The Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong 999077, China
5 shared publications Gilbert Laporte
3 shared publicationsKing-Wah Pang
3 shared publications
Latest Publications
- Open Access
- 21 September 2026
- Published by Elsevier in SSRN Electronic Journal
From Sequential to Integrated Learning and Optimization: Estimation Accuracy, Prescription Performance, and Data Size
Contextual stochastic optimization (CSO) uses covariates to guide decisions under uncertainty, yet its finite-sample behavior is still not well understood. We study two common frameworks for CSO: sequential learning and optimization (SLO) and integrated learning and optimization (ILO). We first show that the estimation-prescription mismatch persists even under stringent pairwise accuracy comparison criteria: a more accurate estimate can induce a worse decision. This motivates the prescriptive curve, which tracks the exact expected out-of-sample cost as a function of training data size under SLO or ILO. We construct contextual linear and nonlinear optimization problems in which these curves strictly increase over a finite range of data sizes under both frameworks, even when the hypothesis class is well-specified. Given this counterintuitive finding, we show that one more data observation is not universally beneficial in the finitesample regime. A single finite-precision sample can be compatible with data-generating processes that imply opposite signs for the one-step change in the prescriptive curve when the training data size increases from N to N+1. We next examine jackknife aggregation as a conservative way to use one additional observation: relative to the traditional procedure trained on one fewer observation, it can strictly improve expected prescription performance under some data-generating processes and strictly worsen it under others. Finally, we extend the analysis beyond SLO and ILO: a universally consistent but finite-sample sub-optimal data-driven policy need not admit another universally consistent policy that, with one additional observation, is never worse across all distributions in an ambiguity set and all finite data sizes, and is strictly better for at least one distribution-sample-size pair.
- 21 September 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessFrom Sequential to Integrated Learning and Optimization: Estimation Accuracy, Prescription Performance, and Data Size
Contextual stochastic optimization (CSO) uses covariates to guide decisions under uncertainty, yet its finite-sample behavior is still not well understood. We study two common frameworks for CSO: sequential learning and optimization (SLO) and integrated learning and optimization (ILO). We first show that the estimation-prescription mismatch persists even under stringent pairwise accuracy comparison criteria: a more accurate estimate can induce a worse decision. This motivates the prescriptive curve, which tracks the exact expected out-of-sample cost as a function of training data size under SLO or ILO. We construct contextual linear and nonlinear optimization problems in which these curves strictly increase over a finite range of data sizes under both frameworks, even when the hypothesis class is well-specified. Given this counterintuitive finding, we show that one more data observation is not universally beneficial in the finitesample regime. A single finite-precision sample can be compatible with data-generating processes that imply opposite signs for the one-step change in the prescriptive curve when the training data size increases from N to N+1. We next examine jackknife aggregation as a conservative way to use one additional observation: relative to the traditional procedure trained on one fewer observation, it can strictly improve expected prescription performance under some data-generating processes and strictly worsen it under others. Finally, we extend the analysis beyond SLO and ILO: a universally consistent but finite-sample sub-optimal data-driven policy need not admit another universally consistent policy that, with one additional observation, is never worse across all distributions in an ambiguity set and all finite data sizes, and is strictly better for at least one distribution-sample-size pair.
- 5 February 2026
- Published by Elsevier in Transportation Research Part E: Logistics and Transportation Review
- Vol. 209
Drone scheduling optimization for shore-to-ship delivery and waste recycling
- 5 February 2026
- Published by Elsevier in Transportation Research Part E: Logistics and Transportation Review
- Vol. 209
Drone scheduling optimization for shore-to-ship delivery and waste recycling
- Open Access
- 1 January 2026
- Published by American Institute of Mathematical Sciences (AIMS) in Electronic Research Archive
- Vol. 34 (9) , 6702-6733
An evaluation framework for integrated learning and optimization by revisiting parameters and hyperparameters
Integrated learning and optimization (ILO), an emerging paradigm for contextual optimization also known as "smart predict-then-optimize" or "end-to-end optimization", typically presents major challenges for practical applications in the community of operations research and management science (OR/MS). This study developed a general evaluation framework that systematically assesses the difficulty of implementing ILO for contextual optimization problems. It distinguishes between computationally tractable and challenging problems, offering the corresponding exact and approximate solution strategies to obtain the final prescriptions, respectively. Our main contributions are threefold. First, we identified conceptual ambiguities in traditional definitions of parameters and hyperparameters in machine learning (ML) models, proposing new complexity-based definitions that strengthen the theoretical foundations. Second, building on this, we developed optimization frameworks to identify the optimal ML models and prescriptions under sequential learning and optimization (SLO) and ILO, respectively. The ILO evaluation framework was then established, which is structured around several key questions regarding the specific characteristics of the ML model and the downstream optimization problem. Third, we illustrated the applicability of the proposed evaluation framework through a numerical case study on a stochastic binary decision-making problem and provided a cautious discussion of the observed numerical comparison. Moreover, we analyzed and compared the efficacy and efficiency of SLO and ILO. This study provides valuable insights and practical guidance for implementing ILO in solving real-world problems in OR/MS. Furthermore, it enriches the literature on the comparative effectiveness of SLO and ILO.
- 1 January 2026
- Published by American Institute of Mathematical Sciences (AIMS) in Electronic Research Archive
- Vol. 34 (9) , 6702-6733
Open AccessAn evaluation framework for integrated learning and optimization by revisiting parameters and hyperparameters
Integrated learning and optimization (ILO), an emerging paradigm for contextual optimization also known as "smart predict-then-optimize" or "end-to-end optimization", typically presents major challenges for practical applications in the community of operations research and management science (OR/MS). This study developed a general evaluation framework that systematically assesses the difficulty of implementing ILO for contextual optimization problems. It distinguishes between computationally tractable and challenging problems, offering the corresponding exact and approximate solution strategies to obtain the final prescriptions, respectively. Our main contributions are threefold. First, we identified conceptual ambiguities in traditional definitions of parameters and hyperparameters in machine learning (ML) models, proposing new complexity-based definitions that strengthen the theoretical foundations. Second, building on this, we developed optimization frameworks to identify the optimal ML models and prescriptions under sequential learning and optimization (SLO) and ILO, respectively. The ILO evaluation framework was then established, which is structured around several key questions regarding the specific characteristics of the ML model and the downstream optimization problem. Third, we illustrated the applicability of the proposed evaluation framework through a numerical case study on a stochastic binary decision-making problem and provided a cautious discussion of the observed numerical comparison. Moreover, we analyzed and compared the efficacy and efficiency of SLO and ILO. This study provides valuable insights and practical guidance for implementing ILO in solving real-world problems in OR/MS. Furthermore, it enriches the literature on the comparative effectiveness of SLO and ILO.
- Open Access
- 1 January 2026
- Published by American Institute of Mathematical Sciences (AIMS) in Electronic Research Archive
- Vol. 34 (9) , 6011-6042
On the coefficients of prescriptiveness in contextual optimization
This paper focused on the coefficients of prescriptiveness in the field of contextual optimization, which measure the efficacy of a prescriptive analytics method in leveraging contextual information for prescribing data-driven decisions. We identified two key deficiencies of the original coefficient of prescriptiveness proposed in the literature. First, this coefficient is problem-dependent, i.e., it lacks a universal scale: its asymptotic upper bound (less than or equal to 1) varies across problems, making cross-problem comparisons infeasible. Second, this coefficient is data-dependent, i.e., its calculation is contingent on specific training and test datasets, limiting its generalizability. To address the two deficiencies, we proposed two novel coefficients of prescriptiveness in this paper: (ⅰ) the adjusted coefficient of prescriptiveness, which replaces the deterministic perfect-foresight benchmark in the original coefficient with the full-information optimum; and (ⅱ) the expected coefficient of prescriptiveness, derived by taking the expectation over the sampling distribution of the training dataset. Meanwhile, we developed their empirical estimators for practical applications. Moreover, rigorous theoretical analysis established two fundamental properties of these coefficients of prescriptiveness. Through extensive numerical experiments based on the newsvendor problem, we validated the effectiveness of our proposed coefficients of prescriptiveness in addressing the identified deficiencies. By refining the evaluation framework for prescriptive analytics, this work advances the generalizability and applicability of the coefficients of prescriptiveness.
- 1 January 2026
- Published by American Institute of Mathematical Sciences (AIMS) in Electronic Research Archive
- Vol. 34 (9) , 6011-6042
Open AccessOn the coefficients of prescriptiveness in contextual optimization
This paper focused on the coefficients of prescriptiveness in the field of contextual optimization, which measure the efficacy of a prescriptive analytics method in leveraging contextual information for prescribing data-driven decisions. We identified two key deficiencies of the original coefficient of prescriptiveness proposed in the literature. First, this coefficient is problem-dependent, i.e., it lacks a universal scale: its asymptotic upper bound (less than or equal to 1) varies across problems, making cross-problem comparisons infeasible. Second, this coefficient is data-dependent, i.e., its calculation is contingent on specific training and test datasets, limiting its generalizability. To address the two deficiencies, we proposed two novel coefficients of prescriptiveness in this paper: (ⅰ) the adjusted coefficient of prescriptiveness, which replaces the deterministic perfect-foresight benchmark in the original coefficient with the full-information optimum; and (ⅱ) the expected coefficient of prescriptiveness, derived by taking the expectation over the sampling distribution of the training dataset. Meanwhile, we developed their empirical estimators for practical applications. Moreover, rigorous theoretical analysis established two fundamental properties of these coefficients of prescriptiveness. Through extensive numerical experiments based on the newsvendor problem, we validated the effectiveness of our proposed coefficients of prescriptiveness in addressing the identified deficiencies. By refining the evaluation framework for prescriptive analytics, this work advances the generalizability and applicability of the coefficients of prescriptiveness.
- 21 November 2025
- Published by Elsevier in European Journal of Operational Research
- Vol. 330 (1) , 245-256
Multi-output shrunken regression trees
- 21 November 2025
- Published by Elsevier in European Journal of Operational Research
- Vol. 330 (1) , 245-256
Multi-output shrunken regression trees
- 28 August 2025
- Published by Elsevier in Transportation Research Part C: Emerging Technologies
- Vol. 180
Equitable port state control in maritime transportation: A data-driven optimization approach
- 28 August 2025
- Published by Elsevier in Transportation Research Part C: Emerging Technologies
- Vol. 180
Equitable port state control in maritime transportation: A data-driven optimization approach
- 16 July 2025
- Published by Elsevier in Computers & Industrial Engineering
- Vol. 208
Optimizing carbon emission allocation for liner shipping companies
- 16 July 2025
- Published by Elsevier in Computers & Industrial Engineering
- Vol. 208
Optimizing carbon emission allocation for liner shipping companies
- Open Access
- 19 June 2025
- Published by MDPI AG in Mathematics
- Vol. 13 (12) , 2033
A Surrogate Piecewise Linear Loss Function for Contextual Stochastic Linear Programs in Transport
Accurate decision making under uncertainty for transport problems often requires predicting unknown parameters from contextual information. Traditional two-stage frameworks separate prediction and optimization, which can lead to suboptimal decisions, as minimizing prediction error does not necessarily minimize decision loss. To address this limitation, inspired by the smart predict-then-optimize framework, we introduce a novel tunable piecewise linear loss function (PLLF). Rather than directly incorporating decision loss into the learning objective based on specific problem, PLLF serves as a general feedback mechanism that guides the prediction model based on the structure and sensitivity of the downstream optimization task. This design enables the training process to prioritize predictions that are more decision-relevant. We further develop a heuristic parameter search strategy that adapts PLLF using validation data, enhancing its generalizability across different data settings. We test our method with a binary route selection task—the simplest setting to isolate and assess the impact of our modeling approach on decision quality. Experiments across multiple machine learning models demonstrate consistent improvements in decision quality, with neural networks showing the most significant gains—improving decision outcomes in 36 out of 45 cases. These results highlight the potential of our framework to enhance decision-making processes that rely on predictive insights in transportation systems, particularly in routing, scheduling, and resource allocation problems where uncertainty plays a critical role. Overall, our approach offers a practical and scalable solution for integrating prediction and optimization in real-world transport applications.
- 19 June 2025
- Published by MDPI AG in Mathematics
- Vol. 13 (12) , 2033
Open AccessA Surrogate Piecewise Linear Loss Function for Contextual Stochastic Linear Programs in Transport
Accurate decision making under uncertainty for transport problems often requires predicting unknown parameters from contextual information. Traditional two-stage frameworks separate prediction and optimization, which can lead to suboptimal decisions, as minimizing prediction error does not necessarily minimize decision loss. To address this limitation, inspired by the smart predict-then-optimize framework, we introduce a novel tunable piecewise linear loss function (PLLF). Rather than directly incorporating decision loss into the learning objective based on specific problem, PLLF serves as a general feedback mechanism that guides the prediction model based on the structure and sensitivity of the downstream optimization task. This design enables the training process to prioritize predictions that are more decision-relevant. We further develop a heuristic parameter search strategy that adapts PLLF using validation data, enhancing its generalizability across different data settings. We test our method with a binary route selection task—the simplest setting to isolate and assess the impact of our modeling approach on decision quality. Experiments across multiple machine learning models demonstrate consistent improvements in decision quality, with neural networks showing the most significant gains—improving decision outcomes in 36 out of 45 cases. These results highlight the potential of our framework to enhance decision-making processes that rely on predictive insights in transportation systems, particularly in routing, scheduling, and resource allocation problems where uncertainty plays a critical role. Overall, our approach offers a practical and scalable solution for integrating prediction and optimization in real-world transport applications.
- Open Access
- 16 June 2025
- Published by MDPI AG in Mathematics
- Vol. 13 (12) , 1987
Improved Fuel Consumption Estimation for Sailing Speed Optimization: Eliminating Log Transformation Bias
Sailing Speed Optimization (SSO) is a crucial problem in shipping operations management, aiming to reduce both operating costs and carbon dioxide emissions. The ship’s sailing speed directly impacts fuel consumption, where fuel consumption is generally assumed to follow a power function with respect to sailing speed. Previous studies have used transformation-based fitting methods, such as logarithmic transformations, to investigate the relationship between sailing speed and fuel consumption using historical data. However, these methods introduce estimation bias and heteroskedasticity, violating the core assumptions of Ordinary Least Squares (OLS) used for general linear regression. To address these issues, we propose two novel fitting methods that directly optimize the original nonlinear model without relying on transformations. By analyzing the characteristics of the objective function, we derive parameter constraints and integrate them into a discrete optimization framework, resulting in improved fitting accuracy. Our methods are validated through extensive case studies, demonstrating their effectiveness in enhancing the reliability of SSO decisions. These methods offer a practical approach to optimizing fuel consumption in real-world maritime operations.
- 16 June 2025
- Published by MDPI AG in Mathematics
- Vol. 13 (12) , 1987
Open AccessImproved Fuel Consumption Estimation for Sailing Speed Optimization: Eliminating Log Transformation Bias
Sailing Speed Optimization (SSO) is a crucial problem in shipping operations management, aiming to reduce both operating costs and carbon dioxide emissions. The ship’s sailing speed directly impacts fuel consumption, where fuel consumption is generally assumed to follow a power function with respect to sailing speed. Previous studies have used transformation-based fitting methods, such as logarithmic transformations, to investigate the relationship between sailing speed and fuel consumption using historical data. However, these methods introduce estimation bias and heteroskedasticity, violating the core assumptions of Ordinary Least Squares (OLS) used for general linear regression. To address these issues, we propose two novel fitting methods that directly optimize the original nonlinear model without relying on transformations. By analyzing the characteristics of the objective function, we derive parameter constraints and integrate them into a discrete optimization framework, resulting in improved fitting accuracy. Our methods are validated through extensive case studies, demonstrating their effectiveness in enhancing the reliability of SSO decisions. These methods offer a practical approach to optimizing fuel consumption in real-world maritime operations.
- 4 June 2025
- Published by Elsevier in Transportation Research Part B: Methodological
- Vol. 198
Data-driven optimization for container ship bunkering management under fuel price uncertainty
- 4 June 2025
- Published by Elsevier in Transportation Research Part B: Methodological
- Vol. 198
Data-driven optimization for container ship bunkering management under fuel price uncertainty