Exploring the optimal factors influencing teachers’ instructional practices performance using machine learning approaches

Effective teaching practices are essential for enhancing classroom instruction and student learning. However, a comprehensive understanding of the factors influencing teachers’ instructional practices remains elusive. This study employs machine learning (ML) to identify the most significant predictors influencing the instructional practices of secondary school teachers in Shanghai, China. Using data from the Teaching and Learning International Survey (TALIS) 2018, we identified key factors that distinguish teachers with high and low levels of instructional practices. The study combined data-driven and domain knowledge methods to identify the most influential features, and then analysed their impact on teachers’ instructional practices using the Shapley Additive exPlanations (SHAP) and Gradient Boosting Machine (GBM) interpretative framework. The findings identified 10 key features, such as Small group work (tt3g42g) and Quieten the class for the lesson (tt3g42l), that distinguished teachers with high and low levels of instructional practices. This discovery enhances classroom teaching quality and teacher performance.