Decision Tree Machine Learning Approach for the Performance Prediction of Asphalt Mixes Modified with Waste Tyre Metal Fibre

The Marshall stability and flow of asphalt mixes are key performance indicators of their durability and suitability for use in the pavement industry. Achieving the optimal bitumen content and volumetric properties through mix design is critical and depends on the characteristics of the materials used. Recycling waste materials in asphalt is also vital for promoting environmental sustainability. The development of machine learning models plays a crucial role in predicting the performance of such asphalt mixes. This study explores the use of a machine learning approach to predict the performance of waste tyre metal fibre-modified asphalt mixes. A dataset consisting of 75 experimental data points from various mix proportions was compiled to train and test the model. The study used 60/70 penetration grade bitumen and five modified mixes with waste tyre metal fibre (WTMF) contents of 0%, 0.375%, 0.75%, 1.125%, and 1.5%. Decision tree regression was effectively employed to establish the relationship between the input variables. The predictive ability of the model was assessed using R-squared, adjusted R-squared, and mean absolute error. The input parameters included fibre content, bitumen content, aggregate percentage, and porosity. Analysis of the input variables showed that stability decreased while flow increased with higher fibre and bitumen contents. With an R² of 0.901 for training and 0.937 for testing phases, decision tree regression proved to be an effective model for predicting the performance of these modified asphalt mixes.