Wire arc additive manufacturing (WAAM) is a method of additive manufacturing (AM), which deposits material layer by layer to build up a wall by melting a wire using arc. The present study tries to investigated the accuracy of surface roughness (Ra) by machine learning classification-based models of stainless-steel structure fabricated by a pulse MIG. The three input parameters (open circuit voltage, wire feed rate, and travel speed) were considered for measuring the Ra, as well as to generate the ML models. For training and testing, experimental datasets are divided with a 80:20 ratio and two ML models namely, K-nearest neighbors (KNN), and Random Forest (RF) are used. Both ML models’ performance was assessed by the value of mean square error (MSE) and percentage of regression of coefficient (R2). The KNN model exhibited the highest accuracy of 99.35% and the MSE is 0.014. Therefore, these models predicted data was used to NSGAII technique for optimizing the parameters. Moreover, the Pareto chart was plotted for understanding the behaviour of parameters.