Survival Analysis Techniques for Identifying Student Dropout: A Systematic Literature Review

Abstract

This work presents a Systematic Literature Review (SLR) on the use of Survival Analysis (SA) techniques to predict student dropout in the context of Software Engineering. Software Engineering education faces high attrition rates, highlighting the need for predictive approaches that support student retention. SA, successfully applied in fields such as healthcare, enables modeling both the occurrence and timing of an event, offering advantages over other methods. Following Kitchenham’s guidelines and the Quasi-Gold Standard approach, 36 primary studies published between 2013 and 2025 were identified. The results show that Cox proportional hazards model is the most widely used technique, along with Kaplan-Meier and other regression approaches. In addition, 29 relevant variables and 16 validation metrics were identified. The review demonstrates that SA facilitates the early detection of at-risk students, although it faces challenges such as the need for extensive datasets and its limited adoption in the educational domain.