Innovation of autonomous learning mode for college english under IoT integrated software platform

Abstract Innovative learning models have evolved due to the growing integration of the IoT with educational technology. Based on a software platform linked with the IoT, this work proposes the concept of an autonomous learning mode of college English courses. By providing individualized, self-paced learning opportunities, the suggested approach seeks to overcome the drawbacks of conventional classroom instruction. To develop and implement an autonomous learning model for college English courses using a software platform integrated with IoT technology, enhancing self-directed learning, personalized instruction, and overall learning outcomes in English language education. The dataset comprises IoT data, edge computing data, student data, learning sessions, and performance metrics, offering real-time insights into learning outcomes and environmental conditions. The data pre-processing model involves cleaning and standardizing the data collected from IoT devices to ensure consistency. This study optimizes predictive modeling using the innovative WSFLA-INT-LR and extracts features using (PCA) Principal Component Analysis. Classifying student performance is improved by integrating WSFLA with INT-LR. The experimental evaluation of the proposed framework is conducted using an offline educational dataset to validate the predictive capability of the WSFLA-INT-LR model. The obtained predictions demonstrate the potential of the framework to support adaptive learning strategies such as personalized content recommendation, learning pace adjustment, and targeted feedback in an IoT-enabled learning platform. The system achieved an accuracy of 98%, precision of 95%, and sensitivity of 95.7%, indicating strong overall effectiveness and reliability. It scored a balanced F1-score of 94% and a high specificity of 96%. The engagement level, measured through learner interaction and completion of assigned learning activities, reached 78%, while students’ performance improvement, calculated from the increase between pre-session and post-session assessment scores, achieved 87%. The system stability of 93% indicates consistent predictive performance across evaluation runs, and the accuracy loss of 0.06% represents the difference between training and testing accuracy, demonstrating strong generalization capability. The proposed model is adaptable to meet the needs of College English Learners. It promotes independent learning and increases student participation, thus enhancing their learning experience.