Data-Driven Online Academic Forecasting Development: – Current Hot Topics Analysis and Future Research Trends

Abstract

Massive amounts of data have emerged in the age of intelligence, facilitating the mining of learners' performance in the classroom, and determining the occurrence of course success in combination with learner behavior data. The keyword co-occurrence, time zone and emergent graphs are plotted by Citespace to explore the research hotspots in the field of academic prediction, and it is found that the machine learning method for academic prediction is still the prevalent prediction method nowadays. The future development of academic prediction is promising in terms of data selection towards multimodality, deep mining towards multi-algorithm integration, integration of academic prediction into implicit data, and construction of hybrid course prediction models.