A bibliometric analysis of artificial intelligence based predictive analytics in education

Educational predictive analysis (EPA), as a cutting-edge direction of educational data mining (EDM), is promoting precise educational decision-making driven by emerging technologies such as big data and artificial intelligence (AI). However, systematic assessments of the global research landscape and development trends in the field remain insufficient. This study uses bibliometric methods to analyze EPA-related literature in the Web of Science Core Collection (WoSCC) from 2015 to 2025, systematically revealing its research growth trajectory, the distribution of core authors and institutions, international collaboration networks, and thematic evolution. Results indicate that EPA research output continues to grow, with China, India, and the United States occupying a central position in the global academic network. Thematic clustering reveals that EPA research is expanding from traditional student performance prediction to multimodal predictive models that integrate behavioral analysis, learning engagement, and educational outcomes, demonstrating broad application prospects in intelligent education systems and the design of personalized learning pathways. Further qualitative analysis results indicate that three core areas are expected to be the focus of future EPA research, namely multimodal data fusion, explainable AI and non-cognitive ability prediction. This study not only provides quantitative evidence for understanding the academic ecology of EPA, but also offers theoretical reference and practical guidance for future research and educational policy making.