Comparative Analysis of Prediction Models for Surface Roughness of Abrasive Water Jet Machined Aluminum <i>In Situ</i> Composites

In situ fabricated Al7075-TiB 2 Metal Matrix Composites (MMCs) were prepared, and machined through Abrasive Water Jet Machining (AWJM) to investigate their machinability and surface roughness prediction. In order to optimize machining parameters, which are nozzle speed (100–500 mm/min), stand-off distance (0.5–2.5 mm), abrasive flow rate (100–300 g/min), and jet pressure (2000–4000 bars), Taguchi’s L25 orthogonal array is used. The ANOVA analysis showed that nozzle speed was the most significant factor affecting surface roughness, accounting for 70% of the variability. Predictive modeling techniques, such as ANN, KNN, and linear regression, were used to predict surface roughness. The KNN model performed better, with an [Formula: see text] 2 value of 0.9069, MSE of 0.02483, and a prediction error of 3.5%. ANN and regression models with prediction errors of 6.8% and 5.2%, respectively, followed. This work presents KNN precision in machining performance modeling to optimize manufacturing processes in search of better surface quality and lower costs.