Abstract:

Background:
Dropout rates among university students continue to be a major source of concern among higher education administrators. Dropout reduces cost efficiency and harms the institution's image. As a result, it is critical to identify the features that contribute to student dropout.

Purpose:
To select suitable features of student dropout using administrative university data.

Methods:
Machine learning methods were used to pre-process data and select relevant features that contribute to student dropout.

Findings:
The study found that the main features to student dropout are participation average mark, number of modules registered, and modules failed.

Conclusions:
Participation average mark, number of modules registered, and modules failed, put students at risk of dropping out either voluntarily or through academic dismissal. It is therefore, up to the university to devise appropriate intervention methods to assist students who are on the verge of dropping out. At risk students can be retained with the right intervention and support.

Keywords:

Machine learning, imbalanced data, student dropout, feature selection.