To what extent do mark expectations of first years align with their academic achievement within biology based degrees at a South African university?

The drop-out rate in South African universities is of concern because it is is costing the government substantial amounts of funding. There are historical reasons why there is a high drop-out rate, these reasons stem from the Apartheid era and how it had an impact on our education system. In this study we aim to shed light on whether the marks that students anticipate achieving are aligned to the marks that they obtain on assessments. Examining this link could shed light on whether first-year students have realistic views of the academic demands at university. This is done by training 6 different models namely Bagging, Random Forest, Decision table, Logistic Regression, Naive Bayes and Multi-Layer Perceptron. All the models are trained using a K-Fold cross validation K=5 to achieve a 80:20 split of training and test data. The best performing model is the Logistic Regression model with 0.692982 accuracy but with the problem in had it is found that the best model to use is the Decision table due to it’s ability to classify students that failed the year with a recall of 0.719 for class ”Fail”.