Machine Learning-based Classification of Higher Education Admission Success for Informed Decision-Making
Year: 2025
Type: conference-paper
Abstract: Today, the rapid development of intelligent information and communication technologies is a crucial and significant aspect of various fields, including higher education. The complexity of decision-making conditions, as well as the volume of data to be processed during the admission, make it difficult to model and analyze the necessary information for both applicants and the Higher Education Institution itself. The paper provides theoretical and experimental comparison of the effectiveness of existing machine learning methods and models for solving the task of predicting the success of admission to a Higher Education Institution, in order to support applicants’ decision-making. The authors investigated the effectiveness of linear and nonlinear machine learning methods, Neural Networks, and ensemble models based on them to solve the binary classification task. Experimental studies were conducted on a cleaned and normalized real-world dataset from Ukrainian Higher Education Institution. Based on the F1-measure, all methods demonstrated high classification accuracy. The highest accuracy, 98%, was achieved by the method involving the combined use of Probabilistic Neural Network and a stacking ensemble of four Support Vector Machines with different kernels (RBF, Sigmoid, Polynomial, and Linear), in comparison with other methods. The results obtained indicate the potential of using such methods as the foundation for intelligent components of information systems to support applicants’ decision-making in Higher Education Institutions.
Source: Procedia Computer Science
Authors: Khrystyna Zub, Michal Gregus
Language: English
Subfield: Computer Science Applications
Field: Computer Science
Domain: Physical Sciences
SDG: Quality education
Open Access status: diamond
OpenAlex ID: W4416640787