Student Performance Evaluation Technique By Applying Support Vector Classification And Metaheuristic Algorithms On The SVC Model's Reliability

In today's academic landscape, institutions face challenges categorizing individuals by skills, anticipating student performance, and improving test outcomes. Early guidance for students is paramount, directing their efforts towards specific areas to boost academic success. This analytical appr...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:淡江理工學刊 Jg. 28; H. 3; S. 653 - 666
Hauptverfasser: Hai Shou, Ying Lu
Format: Journal Article
Sprache:Englisch
Veröffentlicht: 淡江大學 01.01.2025
Schlagworte:
Dingo Optimization Algorithm (DOA)
Dwarf Mongoose Optimization Algorithm (DMO)
Early guidance
G2 grading
Machine learning
ISSN:2708-9967
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:In today's academic landscape, institutions face challenges categorizing individuals by skills, anticipating student performance, and improving test outcomes. Early guidance for students is paramount, directing their efforts towards specific areas to boost academic success. This analytical approach enables educational institutions to mitigate failure rates by leveraging students' past performance in relevant courses to predict their results in a particular program. Using state-of-the-art approaches, strategies, and tools to enhance the learning environment is where educational data mining comes in. This new product uses data mining and machine learning approaches to educational data to offer useful tools for comprehending students' learning environments. The paper innovatively integrates novel optimizers, Dingo Optimization Algorithm (DOA) and Dwarf Mongoose Optimization Algorithm (DMO), with Support Vector Classification (SVC), exploring their effectiveness in enhancing predictive capabilities. Focusing on educational contexts, the study improves SVC functionality by showcasing DMO's superior performance compared to other hybrid models. This research provides valuable insights into the intersection of ML and education, contributing to the understanding of categorizing individuals, predicting student performance, and improving academic outcomes in educational settings. The effectiveness of the models was assessed using four widely used metrics: Accuracy, Precision, Recall, and F1-score. DMO proved to be a practical optimizer when coupled with SVC compared to the other hybrid model. SVDM (SVC+DMO) increased Accuracy, Precision, Recall, and F1-score index values of the SVC (0.909, 0.920, 0.909, 0.906) model in G2 grading after all iterations completed to 0.929, 0.931, 0.929, and 0.927 respectively which is also higher than results of SVDO.
ISSN:2708-9967
DOI:10.6180/jase.202503_28(3).0020