{"title":"Differentiated instruction on undergraduate students based on classification and prediction of students performance using PSO-BP neural network","authors":"Bingxin Wang , Huiting Liu , Yin Gu , Xuan Pan","doi":"10.1016/j.learninstruc.2025.102210","DOIUrl":null,"url":null,"abstract":"<div><h3>Background</h3><div>Differentiated instruction (DI) is increasingly recognized as a strategy to meet the diverse learning needs of students by adapting teaching methods to individual characteristics. However, accurate student classification remains a challenge, limiting DI's effectiveness.</div></div><div><h3>Aims</h3><div>This study aims to develop a more precise, data-driven classification method to enhance DI implementation and improve learning outcomes.</div></div><div><h3>Samples</h3><div>The study utilized a dataset of 1386 second-year Mechanical and Automotive Engineering undergraduates for model training and validation. An independent external test set of 150 students from the same discipline, not included in the initial dataset, was employed to empirically evaluate the implementation of DI.</div></div><div><h3>Methods</h3><div>We integrated educational data mining with a Particle Swarm Optimization-Backpropagation (PSO-BP) neural network to classify and predict student performance. Key factors influencing academic performance, including past grades, student engagement, and teacher-student interaction, were analyzed. Pearson correlation and multicollinearity testing were employed to ensure model robustness, and K-fold cross-validation ensured accuracy.</div></div><div><h3>Results</h3><div>The PSO-BP model significantly improved classification accuracy, particularly in 3-Classes and 5-Classes groupings. Students receiving DI based on these refined classifications demonstrated notable improvements in academic performance compared to those classified using traditional methods.</div></div><div><h3>Conclusion</h3><div>This study offers a scalable, data-driven solution for student classification in DI, addressing the limitations of current models and providing a practical framework for educational institutions to tailor learning experiences, ultimately enhancing student outcomes and teaching efficacy.</div></div>","PeriodicalId":48357,"journal":{"name":"Learning and Instruction","volume":"100 ","pages":"Article 102210"},"PeriodicalIF":4.9000,"publicationDate":"2025-08-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Learning and Instruction","FirstCategoryId":"95","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S0959475225001343","RegionNum":1,"RegionCategory":"教育学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"EDUCATION & EDUCATIONAL RESEARCH","Score":null,"Total":0}
引用次数: 0
Abstract
Background
Differentiated instruction (DI) is increasingly recognized as a strategy to meet the diverse learning needs of students by adapting teaching methods to individual characteristics. However, accurate student classification remains a challenge, limiting DI's effectiveness.
Aims
This study aims to develop a more precise, data-driven classification method to enhance DI implementation and improve learning outcomes.
Samples
The study utilized a dataset of 1386 second-year Mechanical and Automotive Engineering undergraduates for model training and validation. An independent external test set of 150 students from the same discipline, not included in the initial dataset, was employed to empirically evaluate the implementation of DI.
Methods
We integrated educational data mining with a Particle Swarm Optimization-Backpropagation (PSO-BP) neural network to classify and predict student performance. Key factors influencing academic performance, including past grades, student engagement, and teacher-student interaction, were analyzed. Pearson correlation and multicollinearity testing were employed to ensure model robustness, and K-fold cross-validation ensured accuracy.
Results
The PSO-BP model significantly improved classification accuracy, particularly in 3-Classes and 5-Classes groupings. Students receiving DI based on these refined classifications demonstrated notable improvements in academic performance compared to those classified using traditional methods.
Conclusion
This study offers a scalable, data-driven solution for student classification in DI, addressing the limitations of current models and providing a practical framework for educational institutions to tailor learning experiences, ultimately enhancing student outcomes and teaching efficacy.
期刊介绍:
As an international, multi-disciplinary, peer-refereed journal, Learning and Instruction provides a platform for the publication of the most advanced scientific research in the areas of learning, development, instruction and teaching. The journal welcomes original empirical investigations. The papers may represent a variety of theoretical perspectives and different methodological approaches. They may refer to any age level, from infants to adults and to a diversity of learning and instructional settings, from laboratory experiments to field studies. The major criteria in the review and the selection process concern the significance of the contribution to the area of learning and instruction, and the rigor of the study.