<p>The primary goal of this research is to develop soft computing techniques for forecasting the initial void ratio and collapse potential of alluvial sand containing various percentages of soil lumps. The collapse potential and initial void ratio were predicted using established supervised machine learning tools such as artificial neural network (ANN), support vector machine (SVM), particle swarm optimization (PSO), and adaptive neuro-fuzzy inference system (ANFIS). Over 5000 data points were collected from sixty double-consolidation tests performed on different percentages of alluvial lump-sand mixture. The data were utilized in soft computing techniques. Basic soil parameters such as coefficient of uniformity, coefficient of curvature, initial dry density, and the difference between sand and clay percentages, along with various consolidation parameters like volume of compressibility, pressure during the oedometer test, and settlement due to every consolidation pressure change were employed in the prediction process. The formulated models were statistically analyzed to predict the performance. Simulation results highlight the dominance of the ANN Model for predicting the collapse potential and initial void ratio compared to other soft computing techniques. The best-performing ANN Model was further validated using laboratory test results performed on undisturbed soil samples, ensuring its applicability through real-world validation. The predictions aligned closely with the experimental results.</p>