Template-Type: ReDIF-Article 1.0 Author-Name: Mohamed Dif El Idrissi Author-X-Name-First: Mohamed Dif El Author-X-Name-Last: Idrissi Author-Name: Abdelkabir Charkaoui Author-X-Name-First: Abdelkabir Author-X-Name-Last: Charkaoui Author-Name: Abdelouahed Echchatbi Author-X-Name-First: Abdelouahed Author-X-Name-Last: Echchatbi Title: Hybrid multi-agent framework for green supply chain management Abstract: Environmental customer collaboration has recently attracted a big attention from researchers and industrial professionals. Many studies show that companies may reach high performance level by considering customer collaboration and environmental regulations. However, literature in the green supply chain management (GSCM) suggests having more structured collaboration and information exchange processes between supply chain partners based on new technologies. For this reason, this work proposes a hybrid solution based on multi-agent systems (MAS) and mixed integer linear programming (MILP) to automate and facilitate the environmental customer collaboration process. The study demonstrates how MAS can be used in the GSCM context to improve communication and reduce complexity. An industrial study case in the automotive spare parts sector is used to demonstrate the applicability of the established MAS model. Journal: Int. J. of Information and Decision Sciences Pages: 32-50 Issue: 1 Volume: 17 Year: 2025 Keywords: green supply chain management; multi-agent systems; supply chain management; customer collaboration; environmental regulation. File-URL: http://www.inderscience.com/link.php?id=144258 File-Format: text/html File-Restriction: Access to full text is restricted to subscribers. Handle: RePEc:ids:ijidsc:v:17:y:2025:i:1:p:32-50 Template-Type: ReDIF-Article 1.0 Author-Name: S. Vijaya Bharathi Author-X-Name-First: S. Vijaya Author-X-Name-Last: Bharathi Author-Name: A. Manikandan Author-X-Name-First: A. Author-X-Name-Last: Manikandan Title: A novel hybrid meta-heuristic-enabled ensemble learning model with deep feature extraction for crop yield prediction with heuristic ensemble yield Abstract: The fundamental goal of this study is to build and use heuristic-based ensemble learning for improved agricultural production prediction. The squirrel tunicate swarm algorithm (STSA), a hybrid squirrel search algorithm (SSA) and tunicate swarm algorithm (TSA), extracts deep features using the optimised convolutional neural network (O-CNN). The datasets for agricultural production prediction are obtained from public sources, and deep features are extracted using an optimised convolutional neural network (O-CNN). Following that, the optimum deep features are exposed to heuristic-based ensemble learning using three distinct classifiers: linear regression (LR), support vector regression (SVR), and long-short-term-memory (LSTM) regression. The suggested STSA is utilised to calibrate the ensemble learning's three classifiers. When comparing the predicted performance of the developed model to that of other procedures, the proposed Heuristic ensemble yield framework beats previous techniques. Journal: Int. J. of Information and Decision Sciences Pages: 1-31 Issue: 1 Volume: 17 Year: 2025 Keywords: novel crop yield prediction; deep feature extraction; optimised convolutional neural network; heuristic-based ensemble learning; squirrel tunicate swarm algorithm. File-URL: http://www.inderscience.com/link.php?id=144259 File-Format: text/html File-Restriction: Access to full text is restricted to subscribers. Handle: RePEc:ids:ijidsc:v:17:y:2025:i:1:p:1-31 Template-Type: ReDIF-Article 1.0 Author-Name: Arvind Kamble Author-X-Name-First: Arvind Author-X-Name-Last: Kamble Author-Name: Virendra S. Malemath Author-X-Name-First: Virendra S. Author-X-Name-Last: Malemath Title: MVRO-based DRNN: multi-verse rider optimisation-based deep recurrent neural network for intrusion detection in latency constrained cyber physical systems Abstract: The cyber attacks on cyber physical system leads to actuation and sensing behaviour, safety risks, and rigorous damages to the physical object. Therefore, in this paper, multi-verse rider optimisation (MVRO)-based deep recurrent neural network (DRNN) is devised for identifying intrusions in latency-constrained cyber physical systems. In the latency-constrained cyber physical system, the process is carried out using three layers, end point layer, cloud layer, and fog layer. Here, the feature extraction process is performed using the water wave-based improved rider optimisation algorithm (WWIROA) for the classification process. The MVRO approach is the combination of the rider optimisation algorithm (ROA), and multi-verse optimiser (MVO). The DRNN classifier is utilised for the intrusion detection process. In addition, the DRNN classifier is trained using the introduced MVRO technique for better performance. Furthermore, the MVRO-based DRNN technique achieves low latency of 19.23 s, high specificity, sensitivity, and accuracy of 0.929, 0.974, and 0.956, respectively. Journal: Int. J. of Information and Decision Sciences Pages: 110-131 Issue: 1 Volume: 17 Year: 2025 Keywords: intrusion detection; cyber physical system; cloud layer; deep recurrent neural network; DRNN; multi-verse optimiser; rider optimisation algorithm; ROA. File-URL: http://www.inderscience.com/link.php?id=144260 File-Format: text/html File-Restriction: Access to full text is restricted to subscribers. Handle: RePEc:ids:ijidsc:v:17:y:2025:i:1:p:110-131 Template-Type: ReDIF-Article 1.0 Author-Name: Deepika Dadasaheb Patil Author-X-Name-First: Deepika Dadasaheb Author-X-Name-Last: Patil Author-Name: T.C. Thanuja Author-X-Name-First: T.C. Author-X-Name-Last: Thanuja Author-Name: Bhuvaneshwari C. Melinamath Author-X-Name-First: Bhuvaneshwari C. Author-X-Name-Last: Melinamath Title: RiCSO-based RiDeep LSTM: rider competitive swarm optimiser enabled rider deep LSTM for air quality prediction Abstract: This paper is for air quality prediction. Here, the time-series data is considered for the effective prediction of air quality. Moreover, missing value imputation is applied in this model to perform pre-processing. The technical indicators are extracted as features for the effectual prediction of air quality. The rider deep long short-term memory (LSTM) is also included for predicting air quality, trained by a developed RCSO algorithm. Moreover, the developed rider competitive swarm optimisation (RCSO) approach is newly devised by incorporating rider optimisation algorithm (ROA) and competitive swarm optimiser (CSO). The performance of the developed air quality prediction model is evaluated using several error metrics. The introduced air quality prediction system obtained a minimum mean square error (MSE) of 0.10, a root mean square error (RMSE) of 0.31, a mean absolute percentage error (MAPE) of 8.34%, and mean absolute scaled error (MASE) of 0.30. The results demonstrated that the developed RCSO-based rider deep LSTM model attained better performance than other techniques. Journal: Int. J. of Information and Decision Sciences Pages: 51-75 Issue: 1 Volume: 17 Year: 2025 Keywords: air quality prediction; competitive swarm optimiser; CSO; rider optimisation algorithm; ROA; rider deep LSTM; triple exponential moving average; TEMA. File-URL: http://www.inderscience.com/link.php?id=144261 File-Format: text/html File-Restriction: Access to full text is restricted to subscribers. Handle: RePEc:ids:ijidsc:v:17:y:2025:i:1:p:51-75 Template-Type: ReDIF-Article 1.0 Author-Name: Mohamed Haykal Ammar Author-X-Name-First: Mohamed Haykal Author-X-Name-Last: Ammar Author-Name: Ezzedine Ben Aissa Author-X-Name-First: Ezzedine Ben Author-X-Name-Last: Aissa Author-Name: Habib Chabchoub Author-X-Name-First: Habib Author-X-Name-Last: Chabchoub Title: Digital traceability system for road crude oil transport: the Tunisian case Abstract: Traceability systems have been the major focus of various works in the literature. The diversity of the studies is explained by the need to propose systems which adapt to the various sectors' constraints, the objectives and the recommendations of the stakeholders. They are also related to the nature and the products 'value or concerned with the services and especially the various activities performed by the various partners and the information to be exchanged among the stakeholders. It is worth noting that the stakeholders insist that this traceability system have two main roles: the alerts' generation to avoid the risks of incidents and the determination of responsibility in the case of an incident. In this work, we proposed the modelling of the different activities related to the crude oil transportation using the UML language aiming at the proposal system. We also introduced and described in detail the proposed prototype. Journal: Int. J. of Information and Decision Sciences Pages: 76-109 Issue: 1 Volume: 17 Year: 2025 Keywords: traceability; modelling; crude oil; road transport. File-URL: http://www.inderscience.com/link.php?id=144262 File-Format: text/html File-Restriction: Access to full text is restricted to subscribers. Handle: RePEc:ids:ijidsc:v:17:y:2025:i:1:p:76-109 Template-Type: ReDIF-Article 1.0 Author-Name: Neethu Mecheri Author-X-Name-First: Neethu Author-X-Name-Last: Mecheri Author-Name: Roopa Jayasingh Jayasingh Author-X-Name-First: Roopa Jayasingh Author-X-Name-Last: Jayasingh Title: A survey on various Alzheimer classification techniques using 3D MRI images: a challenging overview Abstract: This survey presents 50 research papers focussed on various techniques in Alzheimer classification techniques using 3D MRI images, and the categorisation of the techniques is made based on the fusion-based, convolutional neural network (CNN)-based, random forest (RF)-based and support vector machine (SVM)-based approaches. Finally, the analysis is to be promoted in the survey based on the research technique, publication year, employed tools, utilised dataset, performance measures and achievement of the research methodologies towards Alzheimer classification techniques using 3D MRI images. At the end, the research gaps and issues of the techniques for Alzheimer classification techniques using 3D MRI images is to be revealed. Journal: Int. J. of Information and Decision Sciences Pages: 220-236 Issue: 2 Volume: 17 Year: 2025 Keywords: Alzheimer classification; convolutional neural network; CNN; random forest; support vector machine; SVM; fusion. File-URL: http://www.inderscience.com/link.php?id=146702 File-Format: text/html File-Restriction: Access to full text is restricted to subscribers. Handle: RePEc:ids:ijidsc:v:17:y:2025:i:2:p:220-236 Template-Type: ReDIF-Article 1.0 Author-Name: Pragya Pandey Author-X-Name-First: Pragya Author-X-Name-Last: Pandey Author-Name: Kailash Chandra Bandhu Author-X-Name-First: Kailash Chandra Author-X-Name-Last: Bandhu Title: A consumer behaviour assessment using dimension reduction and deep learning classification Abstract: Consumer behaviour assessment is extremely important for online communities to finding out mindset of customer and changes their views about specific products and services. Customers share their experiences with particular goods, and services on channels and social media, empowered by artificial intelligence for consumer knowledge sharing and acquire new information. In this proposed work, a deep learning model has been developed for statistical tests, statistical analysis using correlation and association testing are performed. The ordinary dimension reduction with principal component analysis and module eigenvalues, followed by a second normalisation phase that maximises the coefficient's size using possible values. The keras library was used on the third layer of the deep learning classification hierarchy with the rectified linear unit and sigmoid activation functions. The average F1-score was 98% accurate and according to the statistics, the proposed strategy had an accuracy of 84% and a recall of 100%. Journal: Int. J. of Information and Decision Sciences Pages: 133-149 Issue: 2 Volume: 17 Year: 2025 Keywords: consumer behaviour; artificial intelligence; principal component analysis; PCA; consumer knowledge-sharing; deep learning classification. File-URL: http://www.inderscience.com/link.php?id=146703 File-Format: text/html File-Restriction: Access to full text is restricted to subscribers. Handle: RePEc:ids:ijidsc:v:17:y:2025:i:2:p:133-149 Template-Type: ReDIF-Article 1.0 Author-Name: Saurab Kumar Author-X-Name-First: Saurab Author-X-Name-Last: Kumar Author-Name: Vraj Patel Author-X-Name-First: Vraj Author-X-Name-Last: Patel Author-Name: Joy Mehta Author-X-Name-First: Joy Author-X-Name-Last: Mehta Author-Name: Jaiprakash Verma Author-X-Name-First: Jaiprakash Author-X-Name-Last: Verma Author-Name: Ankit K. Sharma Author-X-Name-First: Ankit K. Author-X-Name-Last: Sharma Title: A comparison of statistical and machine learning models for stock price prediction Abstract: A huge proportion of money around the world is held by the stock markets. It is one of the most pivotal aspects of the financial institutions and experts. Predicting the movements of stock markets can improve decision making for traders. In this paper, data science techniques are employed to predict the 'closing prices' of stocks. Performance of Facebook Prophet (Fb-P), auto-regressive integrated moving average (ARIMA), simple exponential smoothing (SES), vector autoregression (VAR), random forest (RF), linear regression (LR), k-nearest neighbours (k-NN), decision tree (DT), extreme gradient boost (XGB), and long short-term memory (LSTM) are examined on stock price data of companies listed on the Bombay Stock Exchange (BSE). The proposed work concludes that for extended time frames, LR model performs the best followed by LSTM and ARIMA when compared on the same metrics. Journal: Int. J. of Information and Decision Sciences Pages: 168-193 Issue: 2 Volume: 17 Year: 2025 Keywords: deep learning; machine learning; stock price prediction; statistical modelling. File-URL: http://www.inderscience.com/link.php?id=146704 File-Format: text/html File-Restriction: Access to full text is restricted to subscribers. Handle: RePEc:ids:ijidsc:v:17:y:2025:i:2:p:168-193 Template-Type: ReDIF-Article 1.0 Author-Name: Clóvis Santos Author-X-Name-First: Clóvis Author-X-Name-Last: Santos Author-Name: Carina Dorneles Author-X-Name-First: Carina Author-X-Name-Last: Dorneles Title: Predictive data using linear regression in agricultural production Abstract: In agribusiness some challenges are related to generating information for predictability with an acceptable safety accuracy. In this context, data management systems are usually developed to meet only the operational, legal, and regulatory requirements. The gap in functionalities regarding data science creates the opportunity to develop complementary tools such as business intelligence, data warehousing, online analytics, and others. This paper presents an approach to predict possible scenarios from historical harvested crops datasets. We conducted our proposal using a set of government data on harvests in all regions of Brazil in a historical series of 45 years. We have developed a descriptive application for predictive data analysis and information generation for forecasting scenarios in agriculture, using machine learning with a predictive algorithm implemented with linear regression. Objectively, the results show the use of real datasets to generate possible values in crops according to previous seasons. Journal: Int. J. of Information and Decision Sciences Pages: 150-167 Issue: 2 Volume: 17 Year: 2025 Keywords: agribusiness; database; linear regression data extraction; machine learning. File-URL: http://www.inderscience.com/link.php?id=146711 File-Format: text/html File-Restriction: Access to full text is restricted to subscribers. Handle: RePEc:ids:ijidsc:v:17:y:2025:i:2:p:150-167 Template-Type: ReDIF-Article 1.0 Author-Name: Amira Khelil Author-X-Name-First: Amira Author-X-Name-Last: Khelil Title: Managerial practices for speedy strategic decision in multinational firms Abstract: Strategic decision speed (SDS) is essential for top management teams (TMTs) to lead their organisations effectively in the international business world. However, few managers seem to have actually realised how TMTs could reach an efficient strategic decision (SD). In order to address this issue, we utilised SD literature to determine a set of relevant factors, whereby, decisions could be more easily reached. By relying on the SD and resource based view (RBV) theories, we maintain that such factors as centralisation, enterprise resource planning (ERP), collaborative culture, and intuition are key elements which could help TMTs make prompt decisions and achieve international performance. In this context, using partial least square (PLS) software-based data sources, this study on Tunisia-based multinational firms, indicates that collaborative culture and ERP factors appear to be key antecedents of prompt SD, a significantly influential factor for maintaining international performance. Journal: Int. J. of Information and Decision Sciences Pages: 194-219 Issue: 2 Volume: 17 Year: 2025 Keywords: strategic decision speed; SDS; ERP; collaborative culture; intuition; international performance. File-URL: http://www.inderscience.com/link.php?id=146712 File-Format: text/html File-Restriction: Access to full text is restricted to subscribers. Handle: RePEc:ids:ijidsc:v:17:y:2025:i:2:p:194-219 Template-Type: ReDIF-Article 1.0 Author-Name: David McLain Author-X-Name-First: David Author-X-Name-Last: McLain Author-Name: Jinpei Wu Author-X-Name-First: Jinpei Author-X-Name-Last: Wu Title: The influence of information sources on process and content confidence when making Ill-structured managerial decisions Abstract: Although information is an important influence on decision confidence, disparate views exist about that influence. Engineering-derived psychological theory associates information non-negatively with confidence whereas the overconfidence literature suggests information has a non-positive influence on confidence. Previous research, however, has used information sources lacking ecological validity and almost exclusively studied well-structured decisions and single facets of confidence. Drawing on research and practice in management, decision making, and the neurosciences, the influences of technology-sourced (web) information and performance feedback information were studied as influences on confidence when making ill-structured decisions. Clear distinctions were made between judgment leading up to a decision, called process confidence, and evaluation of the final decision, called content confidence. Process-integrated web use only weakly increased either confidence while feedback significantly reduced content confidence and left process confidence little changed. This effect was amplified when the feedback clarified others' decision expectations. Within subjects, the relationship between quantified feedback and confidence, especially process confidence, was positive and increased with each decision. These findings suggest that an information resource when making ill-structured decisions has little effect on confidence but that credible, post-decision performance information can affect content confidence while process confidence remains resilient. Journal: Int. J. of Information and Decision Sciences Pages: 261-282 Issue: 3 Volume: 17 Year: 2025 Keywords: ambiguity; confidence; decision making; feedback; ill-structured decisions; information; internet; meta-cognition; web. File-URL: http://www.inderscience.com/link.php?id=149300 File-Format: text/html File-Restriction: Access to full text is restricted to subscribers. Handle: RePEc:ids:ijidsc:v:17:y:2025:i:3:p:261-282 Template-Type: ReDIF-Article 1.0 Author-Name: Rohit Kenge Author-X-Name-First: Rohit Author-X-Name-Last: Kenge Title: Sales order booking process lead time reduction by deployment of the lean principle Abstract: Changing environment, government imbalances, and COVID-19 disease are resulting in higher product delivery lead time and longer sales order booking process. Some gaps are identified in the sales order booking process from the current literature survey, and they are addressed by a proposed hypothesis model to reduce the lead time of the sales order booking process. A survey of buyers and salespersons is executed with a set of questionnaires for the period of December 2020 to March 2021 on a convenience sampling basis for the 500 samples by circulating the Google Forms. Four hundred two that is 80.33% correct responses are received. The conclusion of the survey response data analysis after testing the reliability, validity, correlation matrix, and structural equation modelling is our proposed model is the best fit and improved the service level from approximately 89 to 99% for the sales order booking process of our selected electrical OEM products. Journal: Int. J. of Information and Decision Sciences Pages: 283-307 Issue: 3 Volume: 17 Year: 2025 Keywords: sales management; lean; lead time reduction; sales operation excellence; sales order booking process. File-URL: http://www.inderscience.com/link.php?id=149301 File-Format: text/html File-Restriction: Access to full text is restricted to subscribers. Handle: RePEc:ids:ijidsc:v:17:y:2025:i:3:p:283-307 Template-Type: ReDIF-Article 1.0 Author-Name: Rishika Anand Author-X-Name-First: Rishika Author-X-Name-Last: Anand Author-Name: Meenakshi Saroha Author-X-Name-First: Meenakshi Author-X-Name-Last: Saroha Author-Name: Pooja Gambhir Author-X-Name-First: Pooja Author-X-Name-Last: Gambhir Author-Name: Dimple Sethi Author-X-Name-First: Dimple Author-X-Name-Last: Sethi Title: COVID patients' severity level detection using machine learning approach Abstract: COVID-19 is a contagious disease that is caused by the SARS-CoV-2. This disease originated in Wuhan, China, in 2019, which resulted in a pandemic. This virus is diagnosed using chest computed tomography. Preventive measures like not touching face, maintaining distance, and frequent washing hands are taken care of to reduce disease transmission. There is a vaccine for COVID-19, but it is effective to some extent, whereas fewer hospitals are there for the patients suffering from COVID-19 in India. So, the government needs to admit the patients with the severe infection from COVID-19, and the patients with less severity have to isolate themselves in their homes. In this article, various parameters are considered to detect the severity of the patient suffering from COVID-19. Machine learning techniques are applied to get better accuracy while detecting the severity of the patients. Journal: Int. J. of Information and Decision Sciences Pages: 326-341 Issue: 3 Volume: 17 Year: 2025 Keywords: COVID-19; symptoms of COVID-19; machine learning; the severity of patients. File-URL: http://www.inderscience.com/link.php?id=149302 File-Format: text/html File-Restriction: Access to full text is restricted to subscribers. Handle: RePEc:ids:ijidsc:v:17:y:2025:i:3:p:326-341 Template-Type: ReDIF-Article 1.0 Author-Name: Xubo Zhang Author-X-Name-First: Xubo Author-X-Name-Last: Zhang Author-Name: Yanbin Tu Author-X-Name-First: Yanbin Author-X-Name-Last: Tu Author-Name: KeKe Zhong Author-X-Name-First: KeKe Author-X-Name-Last: Zhong Title: The factors driving buyers to post their online feedback - ordered logistic regression analysis Abstract: Reputation is vital for sellers to survive and grow at online auction marketplaces. Positive feedback ratings from buyers help sellers build such a reputation. In this study, we explore the factors that affect feedback ratings from buyers at online auction marketplaces. We try to identify the factors related to three types of feedback (+, 0, -) posted by buyers. We also study the effects of sellers' first move to post their feedback to buyers on the counter-feedback from buyers. We find that characteristics of sellers and products, selling strategies, auction outcomes, and the first mover strategy are significantly associated with feedback posted from buyers. More specifically, sellers' first move to post their positive feedback to buyers helps them receive positive counter-feedback from buyers. Our study contributes to the literature by exploring the determinants of online feedback posted from buyers and providing empirical evidence on the effects of sellers' first mover strategy. Journal: Int. J. of Information and Decision Sciences Pages: 308-325 Issue: 3 Volume: 17 Year: 2025 Keywords: online auction; buyer's feedback; first mover strategy; marketing analytics. File-URL: http://www.inderscience.com/link.php?id=149304 File-Format: text/html File-Restriction: Access to full text is restricted to subscribers. Handle: RePEc:ids:ijidsc:v:17:y:2025:i:3:p:308-325 Template-Type: ReDIF-Article 1.0 Author-Name: R. Santhosh Kumar Author-X-Name-First: R. Santhosh Author-X-Name-Last: Kumar Author-Name: N. Prakash Author-X-Name-First: N. Author-X-Name-Last: Prakash Title: Design of hybrid SVM job recommender system for the overlapping target classes Abstract: The fresh graduates with no prior experience are struggling to find suitable jobs. The job searching time of the fresh graduate is not reduced. Few researchers used machine learning models for matching the recommended job skill-set with the graduate skill set. If the skill-set of the two jobs is the same, the machine learning algorithms recommend only one job and ignore the other job. To address this problem, we design a hybrid support vector machine job recommendation (HSJR) model. The proposed HSJR model collects the skill set of the graduate and matches it with the current jobs and recommends the most suitable jobs for the graduates. To evaluate the proposed HSJR model, the jobs are recommended for the engineering graduates and the feedback received from the participants. The proposed HSJR model achieves 90% accuracy in the job recommendation. The proposed HSJR model performs better than the traditional job recommender system. Journal: Int. J. of Information and Decision Sciences Pages: 243-260 Issue: 3 Volume: 17 Year: 2025 Keywords: job recommender system; support vector machine; SVM; skill set; career recommendation; recommendation system; recommender system. File-URL: http://www.inderscience.com/link.php?id=149305 File-Format: text/html File-Restriction: Access to full text is restricted to subscribers. Handle: RePEc:ids:ijidsc:v:17:y:2025:i:3:p:243-260 Template-Type: ReDIF-Article 1.0 Author-Name: K.D. Mohana Sundaram Author-X-Name-First: K.D. Mohana Author-X-Name-Last: Sundaram Author-Name: T. Shankar Author-X-Name-First: T. Author-X-Name-Last: Shankar Author-Name: N. Sudhakar Reddy Author-X-Name-First: N. Sudhakar Author-X-Name-Last: Reddy Title: Lemon fruit classification by transfer learning technique: experimental investigation of convolutional neural network Abstract: Before exporting fruits, quality control is extremely important in the fruit industries. The most crucial step in the quality assessment process is to classify the fruit as fresh or spoiled. Convolutional neural network (CNN) is the most recent technology used for classification. Henceforth, in this work, the performance of eight widely used CNNs, namely AlexNet, DenseNet, GoogleNet, Inceptionv-3, MobileNetv-2, ResNet-18, SqueezNet, and VGGNet-19, was evaluated and compared for fruit classification, utilising the Lemon fruit dataset. To classify the lemon fruits into three categories of good-quality, medium-quality, and poor-quality, 1,000 fully connected layers in each CNN were substituted with three fully connected layers. For comparison, all of the CNNs were trained using the Transfer Learning technique with learning rates of 0.1, 0.01, and 0.001. The VGG Net-19 architecture was found to have a validation accuracy of 92.6% for a learning rate of 0.001. Journal: Int. J. of Information and Decision Sciences Pages: 401-409 Issue: 4 Volume: 17 Year: 2025 Keywords: convolutional neural network; CNN; fruit classification; fully connected layers; transfer learning. File-URL: http://www.inderscience.com/link.php?id=150096 File-Format: text/html File-Restriction: Access to full text is restricted to subscribers. Handle: RePEc:ids:ijidsc:v:17:y:2025:i:4:p:401-409 Template-Type: ReDIF-Article 1.0 Author-Name: Archit Dhiman Author-X-Name-First: Archit Author-X-Name-Last: Dhiman Author-Name: Mohit Kumar Author-X-Name-First: Mohit Author-X-Name-Last: Kumar Author-Name: Arun Kumar Yadav Author-X-Name-First: Arun Kumar Author-X-Name-Last: Yadav Author-Name: Divakar Yadav Author-X-Name-First: Divakar Author-X-Name-Last: Yadav Title: Pothole detection and localisation from images using deep learning Abstract: The existence of potholes threatens road safety and contributes to a significant portion of accidents worldwide. It takes a lot of work to constantly patch potholes and keep track of when new ones appear. Our goal in this work is to create a pothole detection system that would make it simpler to accurately detect potholes from images. The system can potentially save human lives and assist the government authorities to fix the potholes. In order to achieve this objective, we first make use of a pre-trained deep learning model (VGG-16) and thereafter, propose a novel convolutional neural network (CNN) model. This work employs a publicly available dataset, Nienaber Potholes 2 (Complex), for experiments. The proposed model provides 98.87% accuracy on pothole classification task in images and outperforms recent state-of-the-art approaches in the literature. Further, since no past work has been done on this dataset to detect bounding boxes for potholes, we use YOLO-v3 and YOLO-v5 to generate bounding box predictions on this dataset and evaluate the results. The bounding box task achieves 83.23% mAP and 87.45% precision. Due to the absence of significant existing results, these results for bounding box prediction may be considered as a benchmark. Journal: Int. J. of Information and Decision Sciences Pages: 357-370 Issue: 4 Volume: 17 Year: 2025 Keywords: pothole detection; pothole; convolutional neural network; CNN; bounding box; you only look once; YOLO; Nienaber. File-URL: http://www.inderscience.com/link.php?id=150097 File-Format: text/html File-Restriction: Access to full text is restricted to subscribers. Handle: RePEc:ids:ijidsc:v:17:y:2025:i:4:p:357-370 Template-Type: ReDIF-Article 1.0 Author-Name: Phillip W. Witt Author-X-Name-First: Phillip W. Author-X-Name-Last: Witt Author-Name: Mohsen Hamidi Author-X-Name-First: Mohsen Author-X-Name-Last: Hamidi Title: A binomial simulation approach to more consistent AHP matrices Abstract: Different methods have been proposed for simulating analytical hierarchy process (AHP) matrices. In this paper, we develop a new method assuming responses for AHP matrices follow binomial distributions. In this paper, we exhibit the method with a small data sample which we use to estimate the binomial parameters by transformation, and then backwards solving the logit Newton-Raphson updating algorithm. A parametric bootstrap sample is then used to compare the simulated results against the actual results from the data. In addition to the new simulation method we discover some interesting findings regarding AHP matrix consistency. Journal: Int. J. of Information and Decision Sciences Pages: 343-356 Issue: 4 Volume: 17 Year: 2025 Keywords: analytic hierarchy process; AHP; simulation; decision-making; group decision-making; decision science; managerial decision-making; consistency; Newton-Raphson; binomial. File-URL: http://www.inderscience.com/link.php?id=150098 File-Format: text/html File-Restriction: Access to full text is restricted to subscribers. Handle: RePEc:ids:ijidsc:v:17:y:2025:i:4:p:343-356 Template-Type: ReDIF-Article 1.0 Author-Name: Amit Hedau Author-X-Name-First: Amit Author-X-Name-Last: Hedau Author-Name: S.V.S. Raja Prasad Author-X-Name-First: S.V.S. Raja Author-X-Name-Last: Prasad Author-Name: Sasikanta Tripathy Author-X-Name-First: Sasikanta Author-X-Name-Last: Tripathy Title: Prediction of return on equity using machine learning algorithms: evidence from India Abstract: The present study analysed and predicted the return on equity using machine learning algorithms from the historical financial data during April 2018-March 2022 for construction firms operating in India. The study considered sampling bias method to consider the listed 172 companies from construction sector, as this sector generates the second largest contribution to the GDP of India. The machine learning algorithms is used to model the regression equation. The results indicate that market capitalisation, sales, return on asset, current ratio, earning per share, promoter holdings and profit after tax significantly influence the return on equity for construction firm in India during the study period. We conclude that out of six classifiers, XGBoost is more accurate (86%) to predict the return on equity of the construction firms in India. Finally, a financial performance prediction tool is developed to predict the results. Journal: Int. J. of Information and Decision Sciences Pages: 410-429 Issue: 4 Volume: 17 Year: 2025 Keywords: return on equity; ROE; construction; India; machine learning; XGBoost. File-URL: http://www.inderscience.com/link.php?id=150099 File-Format: text/html File-Restriction: Access to full text is restricted to subscribers. Handle: RePEc:ids:ijidsc:v:17:y:2025:i:4:p:410-429 Template-Type: ReDIF-Article 1.0 Author-Name: Youyuan Wu Author-X-Name-First: Youyuan Author-X-Name-Last: Wu Author-Name: Wei Chong Choo Author-X-Name-First: Wei Chong Author-X-Name-Last: Choo Author-Name: Bolaji Tunde Matemilola Author-X-Name-First: Bolaji Tunde Author-X-Name-Last: Matemilola Author-Name: Jen Sim Ho Author-X-Name-First: Jen Sim Author-X-Name-Last: Ho Title: A bibliometric analysis of information criteria for forecasting volatility Abstract: Volatility forecasting model selection is an essential issue when making financial decisions, which increasingly focus on modelling, forecasting, and evaluation. However, this area has not yet undergone a systematic analysis in the relevant literature. This paper takes advantage of the VOSviewer and bibliometric techniques to overview the temporal distribution of articles, the corresponding author's countries, the citation network, the co-occurrence, the thematic evolution, and the top of the journal or authors or articles. Content analysis was done to 60 pieces of literature, including their data characteristics, theoretical basis, and practical application, as well as suggestions for potential research directions. Through bibliometric techniques and content analysis, this study provides a thorough overview of the research done in the field of volatility forecasting model selection. The research findings indicate that scientific productivity on the subject is expanding rapidly. New methodologies, such as neural networks, have been introduced, necessitating a broad perspective by the researcher in the evaluation of empirical results. Journal: Int. J. of Information and Decision Sciences Pages: 371-400 Issue: 4 Volume: 17 Year: 2025 Keywords: bibliometric analysis; information criteria; volatility forecasting; model selection. File-URL: http://www.inderscience.com/link.php?id=150100 File-Format: text/html File-Restriction: Access to full text is restricted to subscribers. Handle: RePEc:ids:ijidsc:v:17:y:2025:i:4:p:371-400