Prediction of CBR Values of Natural Soil Reinforced with Water Hyacinth Fibers Using Random Forest Regression

This research examined the effect of adding natural fibers, represented by plant stems and roots, on the physical properties of natural soil. The objective was to improve the quality of the soil for pavement subgrades, the road base layers. Different percentages of fibers (0.25% to 1.25%) were mixed with the soil, and measurements were carried out to calculate dry density, moisture content, and California Bearing Ratio (CBR) when soaked, which reflects the strength of the soil under saturated conditions. The results depicts that the inclusion of fibers in soil increased the dry density of the soil slightly, i.e., the soil was denser, and the moisture content was at the right levels to ensure that the soil was neither too wet nor too dry. The CBR values were greatly improved with fibers, with increased load-carrying capacity. For root fibers, the optimum was 0.75% fiber content with a peak average CBR of approximately 5.28%, while for stem fibers, the optimum was at 1% fiber content with a peak average CBR of approximately 5.02%. Increasing the amount of fibers beyond these points resulted in a reduction in strength, supposedly due to inadequate mixing and the presence of voids in the soil. In addition, to predict CBR values in terms of fiber content, fiber type, dry density, and moisture content, a Random Forest Regression model was developed. The model was highly accurate (R² = 0.94), establishing its usability in predicting soil strength as well as establishing the optimal fiber content for soil reinforcement. In general, the research shows that natural fiber reinforcement enhances soil strength, and thus it is a promising and sustainable method for road construction work.