Аннотация
This study investigates the feasibility of predicting students’ critical thinking levels using machine learning techniques applied to academic and behavioral data. Recognizing critical thinking as a core competency in modern education is yet notoriously difficult to measure directly. This research aims to establish relationship between critical thinking proficiency and quantifiable variables such as an academic performance, extracurricular involvement, and course selection. A dataset comprising 500 anonymized student records was compiled and preprocessed to extract relevant features. Three predictive models—Linear Regression, Decision Tree, and Random Forest Regressor—were trained and evaluated using standard performance metrics. Among the three, the Random Forest model achieved the highest predictive accuracy with an R2 score of 0.84, substantially outperforming the Decision Tree (0.65) and Linear Regression (0.37) models. The results indicate that patterns in students’ course preferences, levels of academic achievement, and engagement in non-academic activities collectively provide meaningful insights into their critical thinking capacity. These findings demonstrate that viability of data-driven frameworks for indirectly assessing cognitive skills and have potential applications in curriculum design, early intervention systems, and educational approach policy development. By leveraging accessible education data, the proposed approach contributes to more scalable, objective, and personalized evaluation strategies within broader domain of learning analytics.