Prediction of Children's Learning Effectiveness Using Fine-Tuned Seagull-Optimized Weighted K-Nearest Neighbour (FSOA-KNN)

Abstract

To predict children's learning effectiveness using data mining (DM) technology, addressing the challenge of unemployment among medium- and relatively low-risk learners. With the rapid expansion of academic institutions, the need for accurate prediction models becomes critical. A novel fine-tuned SeagullOptimized Weighted K-Nearest Neighbor (FSOA-KNN) strategy was proposed to improve the prediction of learning outcomes. The research involved 300 students, and their features were collected and analyzed to assess learning effectiveness. Data preprocessing included min-max normalization to scale features within a defined range, ensuring consistency and reducing bias. Experimental results showed that the FSOA-KNN model achieved an precision of 96.5%, accuracy of 98.7%, F-measure of 95.5%, and recall of 96%. These results demonstrate the model’s effectiveness in forecasting children's learning efficiency and identifying students who require additional guidance or counseling. Additionally, the model's performance was compared with traditional KNN and other optimization methods, demonstrating its superior prediction accuracy.