Data-Driven Teaching Effectiveness Assessment Through Logistic Regression for Enhanced Evaluation Systems and Probabilistic Decision-Making

Abstract

Teaching evaluations provide essential feedback for improving educational quality, yet institutions struggle to efficiently utilize unstructured student comments. This study implements ordinal logistic regression to analyze 4,410 bilingual (English-Filipino) student comments, creating a probabilistic framework for predicting teaching effectiveness across standardized evaluation dimensions. The models achieved predictive performance with AUC values ranging from 0.83 to 0.91, with Knowledge of Subject demonstrating 86.9% accuracy. The system demonstrated 75% reduction in manual analysis time while providing quantified uncertainty measures.