Classification of Aesthetic Education Groups Based on Bayesian Optimization

This paper studies the application of a stochastic forest model based on Bayesian optimization in the classification of aesthetic education groups, utilizing the Kaggle dataset for empirical analysis. The dataset encompasses 12 characteristics, including students' age, gender, ethnicity, parents' education level, study duration, absenteeism, tutoring, sports, music, volunteer service, GPA, and class category, to predict class categories such as science, liberal arts, physical education, and art students. Initially, the data was screened for missing values and outliers, with 70% allocated for training and 30% for testing. Precision, Recall, and F1 Score were employed as evaluation metrics. High classification performance was achieved through Bayesian optimization of the random forest model, yielding Precision=0.88, Recall=0.81, and F1 Score=0.83. To further enhance the model's performance, a feature importance analysis within the random forest framework was conducted, resulting in the exclusion of low-correlation features and retention of only GPA, absenteeism, study duration, and age. This refinement significantly improved the model's performance, achieving Precision=0.90, Recall=0.84, and F1 Score=0.86. Finally, a comparative analysis with SVR and decision tree models revealed that the proposed model exhibited superior performance across Precision, Recall, and F1 Score, thereby validating the effectiveness and superiority of the Bayesian optimization-based random forest model for classifying aesthetic education groups.