Decision Analytics Journal最新文献

{"title":"A deep reinforcement learning approach to quality prediction for production profitability optimization","authors":"Jan Mayer ,&nbsp;Lisa-Marie Wienbrandt ,&nbsp;David Michels ,&nbsp;Roland Jochem","doi":"10.1016/j.dajour.2025.100643","DOIUrl":"10.1016/j.dajour.2025.100643","url":null,"abstract":"<div><div>In high-variance manufacturing environments, the early detection and elimination of defective products is crucial for optimizing resource utilization and increasing profitability. Traditional quality control methods often fail to provide timely insights, especially in processes involving complex, multivariate sensor data. To address this gap, this study explores the use of Deep Reinforcement Learning, specifically Deep Q-Learning, as an early classifier for time series data. In order to facilitate the advantages of this approach, data obtained from semiconductor manufacturing, including multivariate sensor data and a final binary classification (good/bad product), is utilized. The proposed approach enables dynamic decision-making during the production process by modeling it as a Markov Decision Process and leveraging experience replay for stable learning. A novel, cost-sensitive reward function is introduced to account for class imbalance and to balance prediction earliness with accuracy. Five distinct models are optimized using hyperparameter tuning based on different classification metrics, and their performance is evaluated in terms of both predictive and economic outcomes. The model optimized with the F1-Metric achieves the best results, with an accuracy of 87% and a mean prediction time of just 1.26 process steps. Economically, this model results in a 22.4% reduction in production time and the highest profit gains across sensitivity analyses.</div></div>","PeriodicalId":100357,"journal":{"name":"Decision Analytics Journal","volume":"17 ","pages":"Article 100643"},"PeriodicalIF":0.0,"publicationDate":"2025-09-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145222252","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An integrated TOPSIS framework with Full-Range Weight Sensitivity Analysis for robust decision analysis","authors":"Àlex Gaona,&nbsp;Albert Guisasola,&nbsp;Juan Antonio Baeza","doi":"10.1016/j.dajour.2025.100642","DOIUrl":"10.1016/j.dajour.2025.100642","url":null,"abstract":"<div><div>Multi-criteria decision-making (MCDM) is widely used in engineering to assist in the selection of the best alternative according to various criteria. Many MCDM methods rely on fixed weight assignments, limiting their ability to reflect uncertainties or variations in decision-maker preferences. This work shows a novel approach that integrates the Technique for Order Preference by Similarity to Ideal Solution (TOPSIS) with conventional one-at-a-time Weight Sensitivity Analysis (WSA) and, for the first time, introduces the Full-Range Weight Sensitivity Analysis (FRWSA). FRWSA enumerates all admissible weight vectors on a discretized simplex and summarizes robustness as dominance frequency, the share of the simplex for which an alternative ranks first. A wastewater treatment plant (WWTP) case study evaluating four different configurations illustrates the approach. Using FRWSA with step <em>h</em> = 0.05 (10 million weight combinations), configuration A2/O-D dominates 82.80% of the weight simplex, with UCT at 12.17%, A2/O-S at 4.94%, and BARD at 0.09%. Comparing with Monte Carlo sampling, FRWSA provides a deterministic, variance-free baseline: MC–Dirichlet with <span><math><mrow><mi>α</mi><mo>=</mo><mn>0</mn><mo>.</mo><mn>5</mn></mrow></math></span> is nearly identical (JSD <span><math><mo>≈</mo></math></span> 0.001 bits), <span><math><mrow><mi>α</mi><mo>=</mo><mn>1</mn><mo>.</mo><mn>0</mn></mrow></math></span> is close but distinct (JSD <span><math><mo>≈</mo></math></span> 0.007-0.008 bits), and plain MC remains farther (JSD <span><math><mo>≈</mo></math></span> 0.043 bits) over 10⁴ – 10⁷ draws. The framework improves transparency via global coverage and boundary diagnostics and is method-agnostic (replicated with VIKOR in the SI). A reusable MATLAB implementation is provided to facilitate adoption. This integrated analysis supports robust and transparent engineering decisions wherever weight uncertainty matters.</div></div>","PeriodicalId":100357,"journal":{"name":"Decision Analytics Journal","volume":"17 ","pages":"Article 100642"},"PeriodicalIF":0.0,"publicationDate":"2025-09-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145222297","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An adaptive framework for privacy-preserving analytics in federated intrusion detection","authors":"Shwetha Jog ,&nbsp;Damodharan Palaniappan ,&nbsp;M.A. Jabbar","doi":"10.1016/j.dajour.2025.100641","DOIUrl":"10.1016/j.dajour.2025.100641","url":null,"abstract":"<div><div>In this paper, we present an adaptive and energy-efficient differentially private federated learning model for IDS in IoT/IIoT environments. The method combines the Fisher Information Matrix (FIM) based parameter pruning, dynamic privacy parameter scheduling and Fast Fourier Transform (FFT) based privacy budget calculation, that strikes a balance between the data utility and the level of achieved differential privacy during training while keeping model complexity and computational overhead to minimum. The efficacy of the proposed framework is substantiated on Edge-IIoTset, a real-world dataset comprising heterogeneous and multiclass attack scenarios, across centralized and federated configurations under different client counts and simulation settings. Results show 65%–72% parameter pruning at <span><math><mo>&gt;</mo></math></span>95% average accuracy over all attack types, consistent cumulative privacy budgets for varying <span><math><mi>ϵ</mi></math></span>-values and better generalization to non-IID data through an adaptive client selection. This manner provides a scalable privacy IDPS for edge-environment with the limited resources.</div></div>","PeriodicalId":100357,"journal":{"name":"Decision Analytics Journal","volume":"17 ","pages":"Article 100641"},"PeriodicalIF":0.0,"publicationDate":"2025-09-24","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145159970","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A machine learning framework for uplift modeling through customer segmentation","authors":"Paulo Pinheiro ,&nbsp;Luís Cavique","doi":"10.1016/j.dajour.2025.100639","DOIUrl":"10.1016/j.dajour.2025.100639","url":null,"abstract":"<div><div>In uplift modeling, the goal is to identify high-value customers based on persuadable customers, those who make a purchase only if contacted. To achieve this, uplift modeling combines machine learning techniques with causal inference, allowing businesses to refine their customer targeting strategies and focus efforts where they are most profitable. This study proposes a practical and reproducible two-phase procedure for identifying high-value customers. In the first phase, customers are segmented using decision trees, which offer a transparent and data-driven approach to grouping individuals with similar characteristics. This segmentation lays the groundwork for a meaningful interpretation of customer behavior. In the second phase, uplift is calculated for each customer segment by comparing the outcomes of the treatment and control groups. This enables the identification of customer groups with the highest uplift. A real-world use case further illustrates the value and applicability of the proposed method. To validate model performance, the procedure employs established metrics such as the Qini index and Cohen’s kappa, which provide insights into both the effectiveness and reliability of the uplift estimates. This work presents a decoupled procedure for uplift modeling that leverages well-established libraries, fostering transparency and a clear understanding of the analytical process. A key contribution to uplift modeling and causal inference is the use of decision trees for stratification, which enables the creation of meaningful segments and their evaluation through the average treatment effect. By integrating theory with practical implementation, this work offers a comprehensive framework for uplift modeling that bridges academic rigor and business usability.</div></div>","PeriodicalId":100357,"journal":{"name":"Decision Analytics Journal","volume":"17 ","pages":"Article 100639"},"PeriodicalIF":0.0,"publicationDate":"2025-09-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145109928","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An integrated approach to hospital efficiency using machine learning and slacks-based super-efficiency evaluation","authors":"Mariana Baptista de Oliveira ,&nbsp;Miguel Alves Pereira ,&nbsp;José Rui Figueira","doi":"10.1016/j.dajour.2025.100640","DOIUrl":"10.1016/j.dajour.2025.100640","url":null,"abstract":"<div><div>Healthcare systems worldwide face mounting pressures due to increasing demand and constrained resources, reinforcing the need for rigorous efficiency measurement to inform policy and managerial decision-making. Traditional Data Envelopment Analysis (DEA) models, while widely applied, are limited by their lack of predictive capability and sensitivity to sample composition. This study addresses these shortcomings by developing and applying an integrated Super-Efficiency Slacks-Based Measure DEA (SuperSBM-DEA) and machine learning (ML) framework to evaluate and predict the efficiency of Portuguese public hospitals between 2014 and 2023. The analysis reveals that 76.82% of hospital units operated inefficiently, with marked regional disparities that reflect historical differences in investment and capacity. Ten ML algorithms were trained to predict DEA efficiency scores, with XGBoost achieving the best performance (<span><math><msup><mrow><mi>R</mi></mrow><mrow><mn>2</mn></mrow></msup></math></span> = 91.46%, RMSE = 0.0438, MAPE = 6.98%). The proposed SuperSBM-DEA-ML framework enables the simulation of counterfactual efficiency scenarios, offering more realistic, stepwise improvement pathways compared to rigid DEA targets. Beyond its predictive accuracy, the framework provides actionable insights for hospital managers and policymakers by supporting forward-looking, data-driven resource allocation and performance monitoring. While the study illustrates the framework’s practical potential, it emphasises that policy adoption should be accompanied by qualitative validation and stakeholder engagement to ensure contextual feasibility and acceptability.</div></div>","PeriodicalId":100357,"journal":{"name":"Decision Analytics Journal","volume":"17 ","pages":"Article 100640"},"PeriodicalIF":0.0,"publicationDate":"2025-09-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145099350","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A data-driven approach to anomaly prediction in automotive display manufacturing","authors":"Hugo Rocha ,&nbsp;Marta Moreno ,&nbsp;Luís Miguel Matos ,&nbsp;Guilherme Moreira ,&nbsp;André Pilastri ,&nbsp;Paulo Cortez","doi":"10.1016/j.dajour.2025.100637","DOIUrl":"10.1016/j.dajour.2025.100637","url":null,"abstract":"<div><div>This study investigates the Wet Optical Bonding (WOB) process in automotive display manufacturing within an Industry 4.0 framework. The objective is to enable early detection of display defects, such as air bubbles and particle contamination, by leveraging tabular WOB input features available before the curing stage, including glass measurements. The anomaly detection task is approached using a range of machine learning (ML) methods. These include Bayesian optimized binary classifiers such as XGBoost, CatBoost, and Deep Feedforward Neural Network; the Automated ML (AutoML) H2O tool; and Bayesian optimized one-class learners, including Isolation Forest and deep Autoencoders. A large industrial dataset of approximately 64,000 WOB records was used to conduct extensive predictive experiments. The evaluation followed a rigorous protocol with internal 3-fold cross-validation for validation and external 10-fold cross-validation for testing, assessing both predictive accuracy and computational efficiency. The ML models demonstrated strong discriminatory performance while maintaining reasonable computational requirements. In addition, a deployment analysis illustrated the potential for reducing the cost and cycle time of the WOB process. Finally, a sensitivity analysis using explainable artificial intelligence (XAI) techniques was conducted to highlight the relevance and influence of key WOB input features.</div></div>","PeriodicalId":100357,"journal":{"name":"Decision Analytics Journal","volume":"17 ","pages":"Article 100637"},"PeriodicalIF":0.0,"publicationDate":"2025-09-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145121060","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An analytical framework for tourism application selection using neutrosophic decision techniques","authors":"A.H. Alamoodi ,&nbsp;O.S. Albahri ,&nbsp;Salem Garfan ,&nbsp;A.S. Albahri ,&nbsp;Tahsien Al-Quraishi ,&nbsp;A.A. Zaidan ,&nbsp;H.A. AlSattar ,&nbsp;Iman Mohamad Sharaf","doi":"10.1016/j.dajour.2025.100638","DOIUrl":"10.1016/j.dajour.2025.100638","url":null,"abstract":"<div><div>In multicriteria decision-making (MCDM), selecting the best option among a set of alternatives by a committee of decision-makers based on given criteria is crucial. Usually, this task is accomplished by using linguistic terms. Researchers apply fuzzy sets in conjunction with a multi-criteria decision-making (MCDM) approach to translate linguistic terms into equivalent fuzzy numbers. However, different degrees of uncertainty and ambiguity arise during this process, which affects the decision-making results. In this context, a complex neutrosophic fuzzy set (CNFS) is employed due to its notable ability to resolve fuzziness and ambiguity in complex environments. Given the practical features of CNFS, this paper integrates and extends two MCDM methods. The research methodology has two phases. The first method is for development, which involves the weighting approach using the developed complex neutrosophic fuzzy-weighted zero-inconsistency (CN-FWZIC) method. This is followed by the second method, called the complex neutrosophic fuzzy decision by opinion score method (CN-FDOSM), which was developed and integrated with CN-FWZIC to prioritize the alternatives. The following main phase included a real-life case study of evaluating and benchmarking tourism data management applications. The results are as follows: (i) The CN-FWZIC method successfully weighs all the smart e-tourism criteria with complete consistency, showing that the highest weight is attributed to the ‘recommender system’ criterion (0.2148540), and the ‘internet of things’ criterion is attributed to the lowest weight, with a score of 0.1057198. (ii) The CN-FDOSM method comprehensively ranks all applications of tourism data management by category. For example, in the ‘tourism marketing’ category, A2 is assigned the first rank with a score of 0.541232539, and A1 is assigned the last rank with a score of 0.314564076. Finally, a robust evaluation was conducted through systematic ranking, sensitivity analysis, complexity analysis, and comparative analysis.</div></div>","PeriodicalId":100357,"journal":{"name":"Decision Analytics Journal","volume":"17 ","pages":"Article 100638"},"PeriodicalIF":0.0,"publicationDate":"2025-09-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145109927","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An analytical framework for real-time gold trading using sentiment and time-series forecasting","authors":"Angel Varela ,&nbsp;Md Kamrul Siam ,&nbsp;Abdullah Al Maruf ,&nbsp;Huanying Gu ,&nbsp;Jerry Q. Cheng ,&nbsp;Zeyar Aung","doi":"10.1016/j.dajour.2025.100633","DOIUrl":"10.1016/j.dajour.2025.100633","url":null,"abstract":"<div><div>Predicting financial markets remains a significant challenge due to their volatility and the complex interplay of various factors influencing price movements. This study presents a method named <em>AchillesV1</em><span><span>¹</span></span>, which is a hybrid long short-term memory (LSTM) neural network model designed to predict the Gold vs. USD exchange rate in real time. The model is trained on high-frequency time-series data and incorporates technical indicators such as the Relative Strength Index (RSI) and Exponential Moving Average (EMA) to enhance prediction accuracy. Additionally, we integrate a FinBERT-based sentiment analysis to evaluate financial news sentiment, further refining trading decisions. The predictions are utilized within an automated trading bot, which executes buy and sell orders based on market conditions. A month-long backtesting experiment (excluding weekends) demonstrated approximately 184% net profit, indicating the model’s effectiveness in live trading scenarios. This work highlights the potential of machine learning in financial markets and contributes to the growing body of research on AI-driven algorithmic trading strategies. Additionally, this trading bot is adaptable to other platforms, including stock and cryptocurrency trading platforms.</div></div>","PeriodicalId":100357,"journal":{"name":"Decision Analytics Journal","volume":"17 ","pages":"Article 100633"},"PeriodicalIF":0.0,"publicationDate":"2025-09-12","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145099351","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An analytical approach to optimizing sustainable farm operations through linear reformulation","authors":"Artur Guerra Rosa ,&nbsp;Pedro Henrique Ferreira Azevedo ,&nbsp;Victor Rafael Rezende Celestino ,&nbsp;Silvia Araújo dos Reis","doi":"10.1016/j.dajour.2025.100632","DOIUrl":"10.1016/j.dajour.2025.100632","url":null,"abstract":"<div><div>The organic food sector has been steadily gaining prominence and expanding its global market share, driving an increasing demand for advanced optimization techniques to enhance the efficiency of sustainable production systems. This paper addresses machinery routing and activity scheduling in a large-scale organic farm case study by developing two mathematical programming decision support models and testing their efficiency. An initial mixed-integer linear programming (MILP) model, inspired by the Traveling Salesman Problem (TSP), was first proposed to optimize farm operations. However, it revealed computational limitations, making the model intractable when scaled to real operational farm demands. To improve efficiency, a linear programming (LP) model based on the previous MILP model was developed to reduce computational complexity and provide flexibility for future integrations. The model performance and scalability were evaluated using resolution time from five different solvers (two commercial and three open-source) across four progressive planning scenarios with scheduling horizons ranging from 7 to 60 days. Results showed that the LP model demonstrates satisfactory efficiency for real-scale farm optimization, achieving timely resolution across all combinations of solvers and planning schedules. Commercial solvers consistently demonstrated the best performance across planning scenarios, while open-source solvers CBC and HiGHS also showed satisfactory solving. Evolving the model proposed from a purely operational tool to a strategic one in the future could align farm logistics with the interconnected goals of the surrounding local food system and community, contributing to Sustainable Development Goals (SDGs) 2 and 12.</div></div>","PeriodicalId":100357,"journal":{"name":"Decision Analytics Journal","volume":"17 ","pages":"Article 100632"},"PeriodicalIF":0.0,"publicationDate":"2025-09-11","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145099348","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A machine learning approach to income inequality from environmental and demographic transitions","authors":"Xiang Qing Lu ,&nbsp;Mingyang Li ,&nbsp;Roengchai Tansuchat ,&nbsp;Woraphon Yamaka","doi":"10.1016/j.dajour.2025.100631","DOIUrl":"10.1016/j.dajour.2025.100631","url":null,"abstract":"<div><div>China’s aging population and green economic transition jointly shape urban–rural​ income disparities in complex ways. This study examines their nonlinear interplay using a hybrid framework that combines traditional econometric models with machine learning techniques, based on panel data from 31 provinces during 2005–2023. Empirical evidence reveals that the green patent ratio (GPR) consistently narrows the income gap, stabilizing disparities when GPR exceeds 10%. Institutional green investment (Green) displays a threshold effect: inequality rises below 0.56% but declines beyond this point, with diminishing marginal returns around 1.5%, suggesting policy saturation. The elderly dependency ratio (old) also shows conditional effects, turning from inequality-reducing to inequality-widening as green investment increases, highlighting resource allocation tensions. Interaction effects, <em>old</em> <span><math><mo>×</mo></math></span> <em>Green</em> and <em>old</em> <span><math><mo>×</mo></math></span> <em>GPR</em>, suggest that effective green development mitigates inequality in aging regions. These findings underscore the importance of coordinated region-specific strategies that integrate demographic trends with green development to promote balanced regional growth in China’s economic transition.</div></div>","PeriodicalId":100357,"journal":{"name":"Decision Analytics Journal","volume":"17 ","pages":"Article 100631"},"PeriodicalIF":0.0,"publicationDate":"2025-09-09","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145099349","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}