"Personalized and Adaptive Machine Learning Framework for Predicting Real-Time Online Learning Comprehension Among Upper Secondary School Students in Gampaha district, Sri Lanka"

"This dataset presents teacher perception data collected to support research on predicting secondary school students\u2019 real-time online learning comprehension in Sri Lanka. The study focuses exclusively on educators\u2019 observations and professional judgments regarding factors that influence student understanding during online learning sessions, including response accuracy, response time, facial expressions, engagement behaviors, and the feasibility of AI-driven predictive systems.Data were obtained from 309 secondary school teachers using a structured questionnaire. A multilayer sampling technique was employed to ensure representative coverage across the Gampaha District, where samples were proportionally selected based on educational zones and the number of schools within each zone. This zonal and school-count\u2013based stratification improved demographic and institutional diversity while minimizing sampling bias.The survey captures teachers\u2019 insights on comprehension indicators, classroom challenges, behavioral cues, demographic influences, and expectations for real-time learning analytics systems. The dataset provides valuable qualitative and quantitative evidence for educational data mining, learning analytics, and adaptive learning research.All data were collected with formal authorization and completed in compliance with ethical standards under the permission of the Sri Lankan Ministry of Education. Personal identifiers were removed to preserve participant anonymity. This dataset supports the development and validation of predictive and intelligent educational support systems for online learning environments. "