Shoukath T. K.
Real-time classroom orchestration requires teachers to infer collective regulatory dynamics from distributed student behavior under severe time constraints, yet existing learning analytics dashboards primarily surface activity metrics rather than modelling the underlying co-regulatory processes that should inform orchestration decisions. This study develops a classroom co-regulation framework based on a tri-variate regulatory state comprising task alignment, behavioral dispersion, and temporal stability derived from 60-second windows of Moodle Learning Management System interaction logs. The framework was evaluated through a seven-layer psychometric validation process involving four-rater expert annotations of 400 windows sampled from a corpus of 1,728 windows collected from undergraduate computing courses at a public university in Oman, with data gathered from 23 course sections over a ten-week period. Results demonstrate strong reliability and construct validity for the proposed state representation, as well as strong agreement among experts regarding instructional actions associated with observed classroom states. Further analysis showed that the derived regulatory indicators systematically differentiated among teaching, independent work, problem-solving, and navigation contexts, supporting their contextual sensitivity and discriminative power. The framework is implementable using standard learning management system export data without supplementary instrumentation, providing a practical foundation for human-in-the-loop classroom orchestration and teacher decision support. The proposed approach contributes a theory-grounded, psychometrically validated representation of classroom co-regulation suitable for real-time educational analytics and future classroom orchestration systems.