An Intelligent Hybrid Machine Learning with Optimization Technique for student performance prediction using online learning activity

Abstract

This instructive center lines up with the acknowledgment that furnishing understudies with pioneering preparing outfits them with the abilities important to make suitable and beneficial answers for arising difficulties. Understanding the elements impacting understudy execution has turned into a pivotal part of schooling research throughout the course of recent many years. Distinguishing these elements adds to further developed understudy execution as well as illuminates showing practices and strategy choices. While internet learning has acquired ubiquity, especially among science understudies, its viability has not generally measured up to assumptions. The developing requests and difficulties in web-based schooling have driven analysts to investigate ways of foreseeing understudy results, including execution and dropout rates in web-based courses. In this review, we propose a wise cross breed AI model with an advancement strategy for anticipating understudy execution in web based learning exercises. This model uses a pre-trained ResNet to extract latent features from the sequence of activities for making predictions. To resolve issues connected with information dimensionality, we present the modified bonobo optimization (MBO) calculation for highlight decrease. Thusly, we utilize an AI procedure, explicitly the deep belief echo state network (DBESN), to foresee understudies' flexibility levels and execution in web based learning exercises. The OULA dataset serves as a validation of the proposed MBO-DBESN method's performance, demonstrating its accuracy in prediction.