Title: Predicting the learning effectiveness of higher mathematics based on learning behaviour analysis

Authors: Fang Huang

Addresses: School of Mathematics, Jiaozuo Normal College, Jiaozuo, 454000, China

Abstract: A prediction method of higher mathematics learning results based on learning behaviour analysis is proposed aiming at the problems of large mean square error and low PL value in the prediction of higher mathematics learning results. First of all, the data of students' learning behaviour is mined and students' files are constructed. Secondly, the dimensionality of learning behaviour data is reduced through the application of the time series t-SNE dimensionality reduction technique, and collaborative filtering algorithm is used to extract students' learning behaviour characteristics. Then, the sample set of learning behaviour characteristics is constructed, and fuzzy clustering algorithm is used for clustering. Finally, the minimum loss function is established by using the random gradient descent method to predict the learning performance based on the federated learning algorithm. The experimental findings indicate that the PL value predicted by this approach consistently stays above 90%, and the model achieves a minimum mean square error (MSE) of 0.05, demonstrating high prediction accuracy and robust performance.

Keywords: learning behaviour; learning effectiveness; student profiles; fuzzy clustering; federated learning.

DOI: 10.1504/IJCEELL.2025.150066

International Journal of Continuing Engineering Education and Life-Long Learning, 2025 Vol.35 No.6, pp.463 - 480

Received: 12 Dec 2024
Accepted: 07 Aug 2025
Published online: 28 Nov 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article