Abstract
Random Forest (RF) is an ensemble learning which creates multiple decision trees and combines their outputs for creating models with less over-fitting. In this study, we apply RF to model the remaining service life (RSL) of rural pavements, a critical factor for developing optimal maintenance strategies and ensuring long-lasting infrastructure. We utilize key variables such as asphalt concrete thickness, base thickness, and surface temperature, along with data from Falling Weight Deflectometer (FWD) measurements. RF demonstrated performance in predicting RSL with consistent accuracy across a variety of conditions. The ensemble nature of RF allows it to effectively manage complex interactions among variables and handle the inherent variability in pavement performance data which makes it well-suited for rural road networks, where environmental and material differences are significant. While some sensitivity to parameter adjustments was noted, the robustness and reliability of RF highlights its potential to be a transformative tool in rural pavement management.