The construction industry is increasingly exploring alternatives to natural aggregates, driven by sustainability concerns and landfill waste reduction. Blast furnace slag, a byproduct of steel manufacturing, exemplifies this shift, serving as a substitute aggregate or concrete additive. This transition supports the circular economy principle, where yesterday’s waste transforms into today’s resources. Key to this practice is the precise determination of material parameters, which vary depending on their origin. Among these, the filtration coefficient is critical, affecting the performance of anthropogenic aggregates in construction and infrastructure. It indicates how well materials transmit water, a factor vital for structural integrity. Machine Learning (ML) presents a promising tool for estimating such parameters efficiently. This paper explores various ML techniques for predicting the filtration coefficient, comparing their effectiveness and examining the impact of the physical properties of aggregates on model accuracy. Through this approach, the paper aims to identify the most suitable methods for parameter estimation, which could enhance the durability and stability of constructions that utilize recycled materials. This research not only contributes to the field of civil engineering but also advances sustainable practices within the industry.