Prognozarea admiterii universitare bazată pe modele statistice şi inteligenţă artificială [Articol]

Abstract

This paper presents a predictive analysis system designed to estimate the future evolution of student enrollment in Moldovan universities. Using official data on natality (1980– 2023), student enrollment, and graduate statistics (2005–2023), the study employs a multi-model approach based on statistical indicators and machine learning forecasting methods. Models such as linear regression, polynomial regressions (degrees 2 and 3), exponential growth, and autoregressive (AR) analysis were implemented and compared. The models are integrated into an interactive Streamlit web application, allowing stakeholders to select forecasting horizons and visualize the future trends of university enrollment by specialty. Results show differentiated evolutions among specialities and highlight the importance of demographic trends in educational planning [1-2].