A hybrid simulation analysis of graduation success at Stellenbosch University

Date
2023-12
Journal Title
Journal ISSN
Volume Title
Publisher
Stellenbosch : Stellenbosch University
Abstract
ENGLISH SUMMARY: Stellenbosch University aims to be one of the leading research institutions on the African continent. High school matriculants (both locally and internationally) can apply to the university with the hope of graduating through one of its ten faculties. The selection of students for enrolment into a degree programme in the Economic Management Sciences Faculty is based on academic merit. For a student to progress to each academic year, they must obtain the minimum academic credits required to continue. The aim of this research is to design and implement a hybrid simulation model of student progression using agent-based and systems dynamics modelling. This model is applied to analyse graduation success in the Faculty. Following the validation of the hybridised student progression model, four intervention scenarios are tested for three degree programmes to increase the number of minimum time graduates. These intervention scenarios focus on a decrease on the influence of the perception of degree difficulty on graduate success, an increase in admission requirements, an increase in student engagement, and a combination approach where all three intervention scenarios are implemented. In the calibrated model for Mathematical Sciences students it results that they are better assisted through higher selection criteria while Management and Economic Sciences students are better assisted through higher engagement. Both the calibrated and uncalibrated models are analysed to control the bias of overfitting. In the non-calibrated model, all students appear to be better assisted through the decreased perception of degree difficulty. The model use is illustrated by means of this specific case study, but it is able to assist decision support in multiple contexts wherever graduation success is a metric of interest.
AFRIKAANSE OPSOMMING: Stellenbosch Universiteit beoog om een van die voorste navorsingsinstansies op die vasteland van Afrika te wees. Hoerskoolmatrikulante (plaaslik sowel as internasionaal) kan by die universiteit aansoek doen met die hoop om deur een van die tien fakulteite te gradueer. Die keuring van studente vir inskrywing vir ’n graadprogram in die Fakulteit Ekonomiese Bestuurswetenskappe is gebaseer op akademiese meriete. Vir ’n student om na elke akademiese jaar te vorder, moet hulle die minimum akademiese krediete verwerf wat nodig is om voort te gaan. Die doel van hierdie navorsing is om ’n hibriede simulasiemodel van studentedeurvloei te ontwerp en te implementeer deur gebruik te maak van agentgebaseerde en stelseldinamiese modellering. Hierdie model word toegepas om gradueringsukses in die Fakulteit te ontleed. Na die validering van die hibriede studentedeurvloeimodel word vier intervensie scenarios op drie graadprogramme getoets om die aantal gegradueerdes binne die minimum tyd te verhoog. Hierdie intervensie scenario’s fokus op ’n afname in die invloed van die persepsie van hoe uitdagend die graadprogram is, ’n toename in toelatingsvereistes, ’n toename in studentebetrokkenheid, en ’n kombinasiebenadering waar al drie intervensie scenarios geimplementeer word. In die gekalibreerde model vir Wiskundige Wetenskappestudente blyk dit dat hulle beter bygestaan word deur hoer keuringskriteria, terwyl Bestuurs- en Ekonomiese Wetenskappestudente beter bygestaan word deur hoer betrokkenheid. Beide die gekalibreerde en ongekalibreerde modelle word ontleed om te beheer vir die vooroordeel van data-oorpassing. In die ongekalibreerde model blyk dit dat alle studente beter gehelp word deur die verminderde persepsie van die moeilikheidsgraad van die kwalifikasies. Die modelgebruik word deur middel van hierdie spesifieke gevallestudie geillustreer, maar is daartoe in staat om besluitnemingsondersteuning in verskeie kontekste te bied, waar gradueringsukses ’n maatstaf van belang is.
Description
Thesis (MCom)--Stellenbosch University, 2023.
Keywords
Citation