Research pick: School’s in - "Utilising a Gaussian process classifier integrating with meta-heuristic optimisers to predict and classify performance systems"

A high-precision approach to predicting university student academic performance and flagging those students at risk of dropping is reported in the International Journal of Internet Manufacturing and Services. The approach combines machine learning with advanced optimisation techniques inspired by natural processes to help universities identify students that are struggling sooner rather than later and allow them to tailor support before problems escalate.

The approach uses a Gaussian Process Classifier (GPC), a statistical model that estimates the likelihood of particular outcomes based on complex, multidimensional data. The GPC is enhanced using an optimisation algorithm inspired by nature, following the way in which organisms locate sources of scent to allow the system to home in on an accurate solution to the question. In addition, it uses a Particle Velocity Search Algorithm (PVSA), based on the collective movement of particles in a fluid. This combination allows the system to fine-tune its parameters to detect subtle patterns in the data regarding student performance, attendance, marks, and engagement levels. In tests, the system could accurately discern which students needed additional support or guidance.

Traditional monitoring methods, while invaluable, often fail to spot early signs of academic difficulty. By contrast, the new model, through the analysis of large datasets, can detect individual behaviour and achievement that change over time and indicate problems earlier.

The work might allow universities to make better data-driven decisions regarding academic support and the distribution of resources, as well as reducing student dropout rates. The researchers suggest that categorising students by learning patterns rather than simple grades could help institutions design more equitable and personalised educational experiences.

Huang, K. and Wang, C. (2025) ‘Utilising a Gaussian process classifier integrating with meta-heuristic optimisers to predict and classify performance systems’, Int. J. Internet Manufacturing and Services, Vol. 11, No. 5, pp.1–30.