Strategic Human Capital Management in the Age of AI: Models for Reducing School Dropout Rates

Abstract

Dropout rates in Angolan higher education represent both a loss of human capital and a real financial problem for institutions. In this article, we analyze the application of machine learning (ML) techniques, specifically K-Prototypes for profile clustering, and Random Forest, SVM, and Decision Tree for early identification of students at risk of dropping out at Kimpa Vita University (UNIKIVI). We used real-world data from 4,354 students across five programs over a seven-year academic period (2016/2017–2022/2023). The models achieved accuracy rates between 93% and 95%, making them useful tools for deciding where to allocate scholarships and psycho-pedagogical support. As for dropout data, the numbers speak for themselves: 16.6% of students left without completing their degree, with the highest concentration in the Accounting and Management, Computer Engineering, and Agronomy programs. We concluded that adopting this type of data-driven governance is not only technically feasible but also economically sensible for all Portuguese-speaking institutions that want to become more financially robust and help train the skilled workforce Angola and Portuguese-speaking countries need.