Annals of Operations Research最新文献

This paper combines two threads of Harry Markowitz’s research—uncertainty and data mining—to demonstrate a methodology for evaluating and improving the accuracy of empirical models and forecasts, focusing on forecasting. Machine learning with indicator saturation provides a generic framework that includes standard techniques for forecast evaluation, such as mean squared forecast errors, forecast encompassing, tests of predictive failure, and tests of bias and efficiency. Saturation techniques are applicable to both economic and non-economic models and forecasts. This paper illustrates the methodology with forecasts of the U.S. federal debt and of the U.S. labor market. Forecast evaluation is fundamental to assess the forecasts’ usefulness and to specify ways in which the forecasts may be improved.

Guerard and Takano (J Investing 1, 48–54, 1992), Guerard et al. (Ann Oper Res 45, 91–108, 1993) and Bloch et al. (Jpn World Econ 5: 3–26, 1993) reported mean–variance efficient portfolios for the Japanese and U.S. equity markets that were composed of a regression-weighted composite model of earnings, book value, cash flow, sales, and their relative variables outperformed their respective equity benchmarks by approximately 400 basis points annually. The optimized portfolios produced higher Sharpe Ratios than the benchmarks in Japan and the United States; the U.S. survivor-biased-free Sharpe Ratio was 1.20 whereas the benchmark was 0.96. Markowitz and Xu (J Portfolio Manag 21, 60–69. 1994) tested the composite model strategy and found that its excess returns were statistically significant from a variety of models tested, and the composite model strategy was not the result of data mining. We report updated US portfolio results for the 1995–2022 period that verifies the Guerard et al. (Ann Oper Res 45, 91–108, 1993) research and demonstrates that the Guerard and Markowitz post-publication, out-of-sample.

Ensemble learning algorithms show good forecasting performances for financial distress in many studies. Despite considering the feature selection and feature importance procedures, most overlook imbalanced data handling. This study proposes the Easyensemble method based on undersampling and combines it with ensemble learning models to predict financial distress. The results show that Easyensemble sampling presents better forecasting performance than SMOTE sampling. We subsequently conduct Permutation Importance (PIMP), Recursive Feature Elimination (RFE), and partial dependence plots, and the experimental results show that the feature selection procedure can effectively reduce the number of indicators without affecting the prediction accuracy, improve the prediction efficiency as well as save processing time. In addition, the indicators from profitability, cash flow, solvency, and structural ratios are essential in predicting financial distress.

This paper examines the aggregate economic performance of 143 countries from 1990 to 2019. Fully nonparametric methods are used to estimate changes in total factor productivity. Recently-developed statistical results are employed for inference and to test hypotheses about model structure. We find that productivity increased by 83 percent on average from 1990 to 2019, with larger gains among low income countries. In addition, we find evidence that productivity growth slowed among high, middle and low-income countries in the 10-year period following the 2008 financial crisis relative to growth over the ten-year period before the crisis. This finding is consistent with what others in the growth literature have found. Finally, our results indicate that low-income countries are less efficient than high- or middle-income countries.

As a service industry, hotel carbon dioxide (CO₂) reduction requires customers’ participation. However, participation in carbon reduction usually involves several specific sacrifices (e.g., inconvenience, lesser quality, or less luxury). Customers may have negative experiences when hotels adopt green practices. Some hotels give up these practices to avoid profit loss risk owing to lower customer satisfaction, which diminishes the effectiveness of CO₂ reduction in the hospitality industry. This paper analytically investigates how to realize a win–win situation that simultaneously reduces CO₂ emissions and increases hotel profit with customer involvement in CO₂ reduction. It is found that seasonal demand, emission costs, CO₂ emission reduction quantity, tourist willingness to involvement, and savings of room operations cost, affect win–win situation realization. This paper also examines the effectiveness of carbon trading, subsidy to low-carbon rooms, increasing public CO₂ reduction awareness, and different hotel low-carbon measure combinations for a win–win situation. Further, insights obtained in this study provide references for hotels on strategy decisions on carbon reduction and for policymakers guiding hotels to achieve low-carbon transformation.

In a highly competitive market environment, outsourcing of logistics service is one of the most significant strategies for enterprises to enhance competitiveness. As a key component of logistics outsourcing, carrier selection directly affects logistics service quality. In general, there are multiple carriers to provide freight service along each route of a logistics network. Each carrier has stochastic capacity owing to that the carrier’s capacity may be partially occupied by other customers. Moreover, goods spoilage often occurs during transportation such that the amount of intact goods delivered may not fulfill the market demand. Thus, each carrier in this study is characterized by stochastic capacity that denotes freight capacity and delivery spoilage that denotes freight quality. Both freight capacity and freight quality are integrated into the carrier selection problem from the viewpoint of network reliability. An effective method that incorporates network reliability algorithm and genetic algorithm is developed to find the optimal carrier selection and the maximum network reliability. A practical logistics network is analyzed to showcase the utility of the algorithm in carrier selection, together with a sensitivity analysis to figure out the important carriers.

Given the growing interest in using AI and analytics to support CRM decision making, we discuss why undersampling and bagging are popular prediction techniques in customer churn prediction (CCP). The former helps in tackling the class imbalance problem and the latter improves model stability. However, extant CCP literature is unclear on the impact of undersampling on model stability and predictive performance, while bagging has difficulties in handling the class imbalance problem. Therefore, we extend existing CCP research to benchmark underbagging, which combines undersampling and bagging. Having both prediction techniques combined we recuperate customer data that would have been lost in undersampling by using them in multiple bags and passing an undersampled, more balanced training set to the classifier. In an extensive experiment including 11 real-life CCP datasets, underbagging is benchmarked against its constituents and other popular CCP classifiers in terms of predictive performance, profit and operational efficiency. Our results indicate that underbagging is a valid and reliable alternative framework for CCP prediction.

We study scheduling problems on parallel dedicated machines and assume that a specific job can only be processed on one specific machine. We concentrate on solving scheduling problems involving convex resource allocation and address three of the most fundamental measures in scheduling theory, i.e., makespan, total load, and total weighted completion time. Firstly, we focus on position-independent workloads, and then we study the setting of general position-dependent workloads, i.e., the workloads are not restricted to be either monotone functions of the job positions or any specific functions. In all problems, we assume a common continuous and non-renewable (limited) resource and adapt known results from scheduling theory to solve the considered problems.

We propose procedures for draws in sports tournaments that will achieve balanced groups of size 3. We start by developing an efficient algorithm that finds all possible balanced groupings; that is, allocations of teams to groups such that the strength of all groups lies between two bounds. The algorithm uses comparison results to discard large sets of unfeasible allocations and backtracking. We then define pot-and-ball draws that randomly pick one of the balanced groupings. Our draw procedures ensure that all groups are of (virtually) identical competitive levels, thereby enhancing the overall fairness and interest of the tournament.