Construction education often faces challenges in providing students with hands-on site experiences due to safety, cost, and accessibility constraints. Mixed Reality Learning Environments (MRLEs) offer a promising solution by simulating real-world construction settings, enabling experiential learning. However, as these environments are increasingly used for training, it is critical to assess and predict users’ Situational Awareness (SA), a key cognitive factor influencing safety and decision-making on construction sites. While prior research has explored SA in MR settings, few have examined predictive models of SA, particularly in the context of implementing sensing technologies like drones and laser scanners for construction education. This study presents a machine learning-based model to predict SA among construction engineering students interacting with a virtual jobsite in MR. Using eye-tracking metrics such as fixation duration, fixation count, and head orientation from construction engineering students wearing the HoloLens, several models were trained to classify SA levels. Among the tested models, a bagging ensemble classifier achieved the highest performance, with a prediction accuracy of 93%. These findings highlight the feasibility of using eye-tracking as a proxy for assessing SA in MR and point toward the development of adaptive learning systems that offer real-time feedback and personalized instruction in construction education. This research contributes to the design of intelligent MR systems that enhance the perception, comprehension, and decision-making of future construction professionals.