This study presents a hybrid framework that combines machine learning techniques with cognitive-behavioural strategies to automatically predict individual student preferences for different types of learning objects and activities in adaptive learning environments. By using behavioural cues and historical interaction data, the model identifies personalised patterns that inform dynamic learning material recommendations, thereby facilitating more engaging and practical educational experiences. Empirical evaluations demonstrate the system’s potential to improve student satisfaction and performance through intelligent adaptive personalisation.

The proposed model is founded on the assumption that student engagement is influenced not only by cognitive factors such as prior knowledge and learning goals but also by behavioural patterns that evolve. By analysing a combination of behavioural signals (e.g., clickstreams, time on task, interaction sequence) and historical interaction data, the system aims to build a rich and contextually relevant learner profile. This profile is then used to inform the selection and sequencing of learning materials that are most likely to resonate with each learner.

To achieve this, the framework employs a multi-layered machine learning architecture that incorporates feature engineering from behavioural data, clustering of learner archetypes, and predictive models to estimate preferences. These models are trained on annotated datasets collected from real-world adaptive learning platforms where learners interact with a variety of media and lesson formats, including videos, readings, simulations, quizzes, and collaborative assignments. Cognitive-behavioural principles are embedded in the model by incorporating metrics related to motivation, effort, and self-regulation, allowing for more nuanced predictions.

Preliminary evaluations show that the hybrid approach outperforms baseline models that rely solely on either behavioural data or static student profiles. User feedback indicates that students perceive the recommendations as more relevant and supportive of their learning paths. This interdisciplinary approach not only enhances the predictive capabilities of adaptive systems but also more closely aligns with the complex, multifaceted nature of human learning.

Keywords: Machine learning, automation, adaptive learning, student's parameters.