Construction of a CatBoost Classification Prediction Model for Municipal-Level Teacher Identity Based on Professional Experience

Abstract

To examine the relationship between teacher identity characteristics and professional experience, the present study uses municipally recognized outstanding teachers in the Beijing-Tianjin-Hebei region as its research sample and develops a teacher identity classification prediction model using the CatBoost machine learning algorithm. Under the guiding premise of “promoting the great spirit of educators,” 14 professional-experience feature indicators are extracted from four dimensions, including educational and instructional competence, collaborative and innovative capacity, and research-and-practice capability, among others. By drawing on CatBoost’s well-established advantage in processing categorical features, the analysis estimates the relative importance of different identity characteristics among outstanding teachers and incorporates those characteristics into variable training and testing procedures. On the methodological front, CatBoost improves on the conventional Gradient Boosting Decision Tree (GBDT): through evaluation on test data, classification performance is sharpened, and model assessment, in turn, is rendered more precise. The findings show that machine learning is broadly applicable to the evaluation of teacher professional development and can accurately disclose the structural composition of teacher professional identity. That, in practical terms, supplies data-based support and scientifically grounded evidence for advancing teacher professional development, designing targeted training strategies, and furthering the construction of a strong education system.