This study was conducted to determine the effect of the STEM approach on increasing student learning activities. The initial hypothesis used is that the STEM approach has an influence on increasing student learning activities. This goal was achieved by conducting a meta-analysis of 20 articles from the ERIC and Google Scholar databases that focused on changes or improvements in student learning activities after the implementation of education with the STEM approach. The results obtained from the analysis using machine learning showed a summary effect value with a percentage of effectiveness of 74%. The summary effect results show that learning with the STEM approach has a positive effect on increasing student learning activities as a whole in several variations of moderate variables used. The involvement of control and experimental classes was carried out to compare the increase in student learning activities using the STEM approach and without using the STEM approach. After a thorough analysis, it was concluded that learning using the STEM approach provides a superior effect compared to without using the STEM approach on increasing student learning activities so as to improve the quality of learning outcomes.