Machine Learning in the Teaching Quality of University Teachers: Systematic Review of the Literature 2014–2024

The growth in the number of students in higher education institutions (HEIs) in Latin America reached 33.5 million in 2021 and more than 220 million worldwide, increasing the number of data volumes in academic management systems. Some of the difficulties that universities face are providing high-quality education to students and developing systems to evaluate the performance of teachers, which encourages offering a better quality of teaching in universities; in this sense, machine learning emerges with great potential in education. This literature review aims to analyze the factors, machine learning algorithms, challenges, and limitations most used to evaluate the quality of teaching based on performance. The methodology used is PRISMA, which considers analyzing literature produced between 2014 and 2024 on factors, prediction algorithms, challenges, and limitations to predict the quality of teaching. Here, 54 articles from journals indexed in the Web of Science and Scopus databases were selected, and 111 factors were identified and categorized into five dimensions: teacher attitude, teaching method, didactic content, teaching effect, and teacher achievements. Regarding the advances in machine learning in predicting teacher teaching quality, 30 ML algorithms were identified, the most used being the Back Propagation (BP) neural network and support vector machines (SVM). The challenges and limitations identified in 14 studies related to HEIs are managing the large volume of data and how to use it to improve the quality of education.