A Multidimensional Evaluation Model of Student Academic Atmosphere Using Random Forest

This study aims to use the Random Forest model to construct and explore a multidimensional evaluation model for students’ academic atmosphere. The study begins with a quantitative analysis of three critical indicators: students’ average normalized grades, frequency of test participation, and grade volatility. Utilizing Principal Component Analysis (PCA), the results reveal that the explained variance ratios are 44.55% for average normalized student grades, 24.72% for the frequency of student test participation, and 30.73% for grade stability. To forge a comprehensive evaluation model for class academic culture, this research further amalgamates various factors such as college affiliation, class attendance rate, overall grades, and class size. The construction of the Random Forest model takes into account the potential influence of class size, making appropriate adjustments. Subsequently, the model is expanded to encompass the college level, aiming to assess and quantify the collective academic ethos of each college. The refined Random Forest model is then applied to the complete student dataset of the school to predict their learning attitudes, integrating these predictions into an expanded dataset. Based on aggregate scores, students’ learning attitudes are categorized into positive and negative.