Decision Analytics Journal最新文献

{"title":"An investigation of supervised machine learning models for predicting drivers’ ethical decisions in autonomous vehicles","authors":"Amandeep Singh,&nbsp;Yovela Murzello,&nbsp;Sushil Pokhrel,&nbsp;Siby Samuel","doi":"10.1016/j.dajour.2025.100548","DOIUrl":"10.1016/j.dajour.2025.100548","url":null,"abstract":"<div><div>Vehicle-pedestrian interactions in autonomous vehicles (AVs) present complex challenges that require advanced decision-making algorithms. Understanding the factors influencing ethical decision-making (EDM) in critical situations is essential as AVs become more prevalent. This study addresses a gap in AV research by using predictive analytics methods to develop models that assess decision-making outcomes under varying time pressures. We recruited 204 participants from North America, aged 18-30 years and 65 years and above, for an online experiment. Participants viewed video clips from a driving simulator that simulated ethical dilemmas. They had to decide whether the AV should stay in its lane or change lanes by pressing the spacebar. The principal component analysis identified age, distraction, and trust in automation as the key factors influencing decision-making. Several machine learning models were optimized to predict decision outcomes, with the Gaussian Naive Bayes model demonstrating strong performance across different time pressures. Feature importance analysis highlighted the significant roles of age and trust in automation. Partial dependence plots illustrated the interaction between these factors and their influence on decision-making outcomes under time constraints. These findings contribute to the development of personalized decision-making algorithms for AVs. Predictive analytics provides valuable insights into improving AV systems’ safety, trust, and ethical behavior by accounting for individual differences in decision-making.</div></div>","PeriodicalId":100357,"journal":{"name":"Decision Analytics Journal","volume":"14 ","pages":"Article 100548"},"PeriodicalIF":0.0,"publicationDate":"2025-01-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143136016","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An outlier detection framework for Air Quality Index prediction using linear and ensemble models","authors":"Pradeep Kumar Dongre ,&nbsp;Viral Patel ,&nbsp;Upendra Bhoi ,&nbsp;Nilesh N. Maltare","doi":"10.1016/j.dajour.2025.100546","DOIUrl":"10.1016/j.dajour.2025.100546","url":null,"abstract":"<div><div>The Air Quality Index (AQI) is a key indicator for assessing air quality and its associated health impacts. Accurate AQI calculations are crucial for reliable air quality assessments, but outliers in air quality data can distort these calculations, leading to inaccurate predictions. This paper presents a comprehensive framework for air quality prediction that integrates multiple outlier detection methods with machine learning models, focusing on enhancing the accuracy and robustness of predictions. The study investigates various outlier detection techniques, including the Interquartile Range (IQR), robust Z-score, and Mahalanobis distance, and evaluates their impact when integrated into machine learning models. Unlike traditional approaches that remove outliers without considering seasonal effects, this research proposes retaining extreme data points after seasonal validation to improve model generalization and prediction accuracy for unseen data. The framework is evaluated using a dataset from Jaipur city, testing multiple machine learning models, including linear regression, ensemble methods, and K-Nearest Neighbor (KNN) regression. Results show that the integrated framework significantly improves model performance, with the Extra Trees Regressor achieving the best results (MAE = 11.9161, RMSE = 16.1660, and <span><math><msup><mrow><mi>R</mi></mrow><mrow><mn>2</mn></mrow></msup></math></span> = 0.8884) after refinement, compared to baseline performance (MAE = 12.6765, RMSE = 17.8452, and <span><math><msup><mrow><mi>R</mi></mrow><mrow><mn>2</mn></mrow></msup></math></span> = 0.8737). This study demonstrates the empirical effectiveness of the proposed framework and provides practical guidelines for air quality prediction in real-world applications.</div></div>","PeriodicalId":100357,"journal":{"name":"Decision Analytics Journal","volume":"14 ","pages":"Article 100546"},"PeriodicalIF":0.0,"publicationDate":"2025-01-16","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143136018","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A systematic review of machine learning applications in sustainable supplier selection","authors":"Joachim O. Gidiagba ,&nbsp;Lagouge K. Tartibu ,&nbsp;Modestus O. Okwu","doi":"10.1016/j.dajour.2025.100547","DOIUrl":"10.1016/j.dajour.2025.100547","url":null,"abstract":"<div><div>Supplier quality evaluation in industrial sectors is well-recognized due to its direct impact on quality assurance and improvement. This task is challenging due to the need to process extensive qualitative and quantitative data, multi-dimensional attributes, and numerous suppliers. Traditional methods are increasingly inadequate to manage this large amount of data and facilitate effective decision-making. This research paper presents a systematic literature review on adopting machine learning techniques for sustainable supplier selection from 2010 to 2024. From an initial pool of 99 papers in the Scopus database, 20 papers from Web of Science, and 52 papers in Google Scholar, 25, 12 and 13 papers, respectively, were selected for in-depth analysis. The study elucidates the role of machine learning in enhancing supplier selection across various sectors, focusing on literature published in English. The findings indicate that machine learning significantly improves organizational performance by refining supplier selection processes, addressing inconsistencies in traditional methods, and leveraging vast data repositories. Integrating Artificial Intelligence (AI) into supply chain operations enables more rapid and reliable decision-making, especially when conventional approaches falter due to large data volumes. The developed framework identifies the significance of traditional and machine learning methods in different stages of supplier selection, including criteria definition, weighting, evaluation, and ranking.</div></div>","PeriodicalId":100357,"journal":{"name":"Decision Analytics Journal","volume":"14 ","pages":"Article 100547"},"PeriodicalIF":0.0,"publicationDate":"2025-01-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143136017","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A cognitive platform for collecting cyber threat intelligence and real-time detection using cloud computing","authors":"Prasasthy Balasubramanian,&nbsp;Sadaf Nazari,&nbsp;Danial Khosh Kholgh,&nbsp;Alireza Mahmoodi,&nbsp;Justin Seby,&nbsp;Panos Kostakos","doi":"10.1016/j.dajour.2025.100545","DOIUrl":"10.1016/j.dajour.2025.100545","url":null,"abstract":"<div><div>The extraction of cyber threat intelligence (CTI) from open sources is a rapidly expanding defensive strategy that enhances the resilience of both Information Technology (IT) and Operational Technology (OT) environments against large-scale cyber-attacks. However, for most organizations, collecting actionable CTI remains both a technical bottleneck and a black box. While previous research has focused on improving individual components of the extraction process, the community lacks open-source platforms for deploying streaming CTI data pipelines in the wild. This study proposes an efficient platform capable of processing compute-intensive data pipelines, based on cloud computing, for real-time detection, collection, and sharing of CTI from various online sources. We developed a prototype platform (TSTEM) with a containerized microservice architecture that uses Tweepy, Scrapy, Terraform, Elasticsearch, Logstash, and Kibana (ELK), Kafka, and Machine Learning Operations (MLOps) to autonomously search, extract, and index indicators of compromise (IOCs) in the wild. Moreover, the provisioning, monitoring, and management of the platform are achieved through infrastructure as code (IaC). Custom focus-crawlers collect web content, processed by a first-level classifier to identify potential IOCs. Relevant content advances to a second level for further examination. State-of-the-art natural language processing (NLP) models are used for classification and entity extraction, enhancing the IOC extraction methodology. Our results indicate these models exhibit high accuracy (exceeding 98%) in classification and extraction tasks, achieving this performance within less than a minute. The system’s effectiveness is due to a finely-tuned IOC extraction method that operates at multiple stages, ensuring precise identification with low false positives.</div></div>","PeriodicalId":100357,"journal":{"name":"Decision Analytics Journal","volume":"14 ","pages":"Article 100545"},"PeriodicalIF":0.0,"publicationDate":"2025-01-08","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143136019","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An integrated multi-stage decision model for upstream supply chain disaster management readiness assessment","authors":"Detcharat Sumrit,&nbsp;Orawan Jongprasittiphol","doi":"10.1016/j.dajour.2024.100538","DOIUrl":"10.1016/j.dajour.2024.100538","url":null,"abstract":"<div><div>Supply Chain Disaster Management (SCDM) is essential for mitigating disruptions, ensuring business continuity, and maintaining a competitive advantage. This study introduces a comprehensive decision model to assess disaster management readiness within the upstream supply chain. Through an extensive literature review grounded in Contingency Resource-Based View (CRBV) theory, fifteen initial readiness factors (<em>RFs</em>) are identified. The Fuzzy Delphi Method (FDM) then refines these factors to twelve, enhancing their relevance and applicability. Next, the Fuzzy Linguistic Preference Relation (FLinPreRa) method assesses the relative importance of each <em>RF</em>, providing a profound understanding of their individual and collective impact. Finally, Weight-Variance Analysis (WVA) evaluates the strengths and weaknesses of each <em>RF</em>, enabling targeted strategies for enhancing disaster readiness. A case study involving five automotive manufacturers in Thailand showcases the practical application of this decision model. The results indicate that this approach is an effective tool for assessing SCDM readiness, pinpointing critical areas for improvement, and guiding strategic investment. Beyond enhancing disaster preparedness, the model also strengthens overall supply chain resilience and responsiveness. Moreover, the framework can be easily adapted to other industries aiming to improve their SCDM readiness by tailoring it to address sector-specific challenges.</div></div>","PeriodicalId":100357,"journal":{"name":"Decision Analytics Journal","volume":"14 ","pages":"Article 100538"},"PeriodicalIF":0.0,"publicationDate":"2025-01-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143136020","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A bi-objective stochastic model for operating room scheduling considering surgeons’ preferences and collaborative surgeries","authors":"Rana Azab ,&nbsp;Amr Eltawil ,&nbsp;Mohamed Gheith","doi":"10.1016/j.dajour.2024.100544","DOIUrl":"10.1016/j.dajour.2024.100544","url":null,"abstract":"<div><div>Operating Rooms (ORs) are pivotal hospital resources, significantly impacting expenses and revenue. This paper introduces a stochastic bi-objective model for OR allocation and scheduling of elective surgeries, considering surgeons’ preferences for specific ORs and preferred start times, as well as the integration of collaborative surgeries (CSs)—where multiple surgeons collaborate to perform a procedure. The proposed stochastic model, which accounts for the inherent uncertainty in surgery durations, seeks to minimize operating costs while maximizing surgeons’ preferences, thus offering a balanced solution for hospital management and medical staff. The model is formulated as a Mixed-Integer Linear Programming (MILP) problem and solved using the Sample Average Approximation (SAA) method. A comprehensive sensitivity analysis was conducted to precisely determine the optimal sample size, defined as the number of scenarios used to model the uncertain surgery durations, to ensure the robustness of the proposed approach. This is essential for approximating the probability distribution of surgery durations, for which a lognormal distribution was employed. This analysis enables stable results concerning the variability in surgery durations. Subsequently, the model was applied to a synthesized dataset, which mirrors real hospital operations. The results demonstrated that the model generates optimal OR and surgeon schedules robust enough to accommodate the inherent variability in surgery durations. Additionally, a Pareto-front analysis was employed to examine the trade-off between minimizing operating costs and maximizing surgeons’ preferences. Implementing a bi-objective optimization algorithm using the <span><math><mi>ɛ</mi></math></span>-constraint method identified a set of optimal schedules, offering valuable insights into balancing cost efficiency and surgeon satisfaction, thereby enabling hospital administrators to make informed scheduling decisions. Extensive numerical experiments were conducted to test the model’s scalability and effectiveness in generating optimal schedules under various operational conditions. The results of these experiments suggest that future work could focus on leveraging heuristic techniques to enhance computational efficiency. In conclusion, the proposed stochastic bi-objective model represents a comprehensive and flexible strategy for enhancing operational efficiency and improving surgeon satisfaction in the allocation and scheduling of ORs.</div></div>","PeriodicalId":100357,"journal":{"name":"Decision Analytics Journal","volume":"14 ","pages":"Article 100544"},"PeriodicalIF":0.0,"publicationDate":"2025-01-04","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143136767","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An integrated mathematical epidemiology and inventory model for high demand and limited supplies under uncertainty","authors":"Yofre H. Garcia ,&nbsp;Saul Diaz-Infante ,&nbsp;Jesus A. Minjarez-Sosa","doi":"10.1016/j.dajour.2024.100543","DOIUrl":"10.1016/j.dajour.2024.100543","url":null,"abstract":"<div><div>At the start of the Coronavirus Disease (COVID-19) vaccination campaign in Mexico, the vaccine was the world’s most essential and scarce asset. Managing its administration to optimize its use was, and still is, of paramount importance. However, when the first vaccine was developed at the end of 2020, due to unprecedented demands and early manufacturing of vaccines, decision-makers had to consider the management of this asset with high uncertainty. We aim to analyze how random fluctuations in reorder points and delivery quantity impact the mitigation of a given outbreak. Because decision-makers would need to understand the implications of planning with a volatile vaccine supply, we have focused our effort on developing numerical tools to evaluate vaccination policies. One of our main objectives is to determine how many vaccines to administer per day so that a hypothetical vaccine inventory keeps its integrity while optimizing the mitigation of the outbreak. Our research uses classic models from inventory management and mathematical epidemiology to quantify uncertainty in a hypothetical vaccine inventory. By plugging a classic inventory model into an epidemic compartmental structure, we formulate a problem of sequential decisions. Then, we investigate how the random fluctuations in the reorder time and number of doses in each delivery impact a hypothetical ongoing vaccine campaign. Our simulations suggest that sometimes, it is better to delay vaccination until the vaccine supply is large enough to achieve a significant response.</div></div>","PeriodicalId":100357,"journal":{"name":"Decision Analytics Journal","volume":"14 ","pages":"Article 100543"},"PeriodicalIF":0.0,"publicationDate":"2025-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143136768","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An efficiency score unification method in data envelopment analysis using slack-based models with application in banking","authors":"Ali Hadi ,&nbsp;Alireza Amirteimoori ,&nbsp;Sohrab Kordrostami ,&nbsp;Saeid Mehrabian","doi":"10.1016/j.dajour.2024.100541","DOIUrl":"10.1016/j.dajour.2024.100541","url":null,"abstract":"<div><div>Providing a unique estimation model of the efficiency score for a bank branch has long been a primary concern for bank managers, who frequently reject efficiency studies because they contend that a single perspective evaluation cannot adequately display the multifunctional nature of decision-making units (DMUs). This paper presents a unification model for efficiency scores in the banking industry. The proposed model evaluates the efficiency of decision-making units through a two-stage method from different perspectives. This study calculates efficiency scores from several aspects and various inputs/outputs in the first stage, so an expanded space is used in the second stage. All viewpoints are transferred into a new space, creating a new efficient frontier in the expanded space. This model presents the unified efficiency score from DMUs by applying the slack-based model (SBM), which proposes enhancement guidelines. A unified efficiency score is created by considering three perspectives: production, profitability, and intermediary. Unlike the average score, the results show that the unified efficiency score can reflect the performance difference between the three scores achieved from the three perspectives. Additionally, this method demonstrates that DMUs cannot achieve overall efficiency if they are inefficient in at least one of the three aspects.</div></div>","PeriodicalId":100357,"journal":{"name":"Decision Analytics Journal","volume":"14 ","pages":"Article 100541"},"PeriodicalIF":0.0,"publicationDate":"2024-12-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143136014","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"An investigation of ensemble learning techniques for obesity risk prediction using lifestyle data","authors":"Shahid Mohammad Ganie ,&nbsp;Bobba Bharath Reddy ,&nbsp;Hemachandran K ,&nbsp;Manjeet Rege","doi":"10.1016/j.dajour.2024.100539","DOIUrl":"10.1016/j.dajour.2024.100539","url":null,"abstract":"<div><div>Obesity disease is a significant health issue and has affected millions of people worldwide. Identifying underlying reasons for the onset of obesity risk in its early stage has become challenging for medical practitioners. The growing volume of lifestyle data related to obesity has made it imperative to employ effective machine-learning algorithms that can gather insights from the underlying data trends and identify critical patient conditions. In this study, an ensemble learning approach including boosting, bagging, and voting techniques was used for obesity risk prediction based on lifestyle dataset. Specifically, XGBoost, Gradient Boosting, and CatBoost models are used for boosting, Bagged Decision Tree, Random Forest, and Extra Tree models are used for bagging, and Logistic Regression, Decision Tree, and Support Vector Machine models are used for voting. Different preprocessing steps were employed to improve the quality assessment of the data. Hyperparameter tuning and feature selection and ranking are also used to achieve better prediction results. The considered models are extensively evaluated and compared using various metrics. Among all the models, XGBoost performed better with an accuracy of 98.10%, precision and recall of 97.50%, f1-score of 96.50%, and AUC-ROC of 100%, respectively. Additionally, weight, height, and age features were identified and ranked as the most significant predictors using the recursive feature elimination method for obesity risk prediction. Our proposed model can be used in the healthcare industry to support healthcare providers in better predicting and detecting multiple stages of obesity diseases.</div></div>","PeriodicalId":100357,"journal":{"name":"Decision Analytics Journal","volume":"14 ","pages":"Article 100539"},"PeriodicalIF":0.0,"publicationDate":"2024-12-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143136015","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"A graph convolutional network for optimal intelligent predictive maintenance of railway tracks","authors":"Saeed MajidiParast ,&nbsp;Rahimeh Neamatian Monemi ,&nbsp;Shahin Gelareh","doi":"10.1016/j.dajour.2024.100542","DOIUrl":"10.1016/j.dajour.2024.100542","url":null,"abstract":"<div><div>This study presents a prescriptive analytics framework for optimal intelligent predictive maintenance of railway tracks. We use machine learning and Graph Convolutional Networks (GCNs) to optimize the maintenance schedules for railway infrastructure and enhance operational efficiency and safety. The model leverages vast data, including geometric measurements and historical maintenance records, to predict potential track failures before occurrence. This proactive maintenance strategy promises to reduce downtime and extend the lifespan of railway assets. Through detailed computational experiments, the effectiveness of the proposed model is demonstrated, providing a significant step forward in applying advanced machine learning techniques to the maintenance of critical transportation infrastructures.</div></div>","PeriodicalId":100357,"journal":{"name":"Decision Analytics Journal","volume":"14 ","pages":"Article 100542"},"PeriodicalIF":0.0,"publicationDate":"2024-12-30","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"143136770","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}