Assessing Economic Vulnerability and Budgetary Effort for Urban Mobility in Bukavu: Insights from Machine Learning Predictions

The city of Bukavu, Democratic Republic of Congo, faces multiple urban mobility challenges. Already confronted with major socio-economic issues, it must cope with rapid population growth, accelerated and unplanned urbanization, and signifi cant defi cienciesin transport infrastructure. Approximately 70% of the population lives in precarious conditions, severely limiting access to reliable transport services. Furthermore, due to its hilly terrain, the city is strongly affected by climate-related disruptions, which negatively impact transport networks and increase user costs. These factors exacerbate economic vulnerability related to mobility and strain household budgets. This study adopts a quantitative, predictive approach to better understand and anticipate household vulnerability in daily mobility, contributing to more inclusive public policy development. The analysis relies on Principal Component Analysis (PCA) to identify key determinants, followed by clustering to segment households according to their vulnerability level. Finally, three predictive models were compared. Cluster analysis revealed three distinct vulnerability profi les, highlighting marked socio-economic stratifi cation. The most disadvantaged households, representing 73% of the sample, are the most exposed, with budget shares reaching up to 50% of income. Conversely, affl uent and middle-class households enjoy better mobility conditions but remain sensitive to economic and climate shocks. These fi ndings underscore the need to integrate spatial and economic inequalities into local public policy planning. Among the tested models, logistic regression stood out for its accuracy and its ability to identify vulnerable households with perfect recall.