It is challenging to model the coefficient of friction, surface roughness, and related tribological processes during metal contact because of flattening, ploughing, and adhesion. It is important to choose the appropriate process parameters carefully when creating analytical models to overcome the challenges posed by complexity. This will ensure the production of sheet metal formed components that meets the required quality standards and is free from faults. This research analyses the impacts of nominal pressure, kinematic viscosity of lubricant, and lubricant pressure on the coefficient of friction and average roughness of DC05 deep-drawing steel sheets. The strip drawing test was used to determine the coefficient of friction. This work utilises the Categoric Boosting (CatBoost) machine learning algorithm created by Yandex to estimate the COF and surface roughness, intending to conduct a comprehensive investigation of process parameters. A Shapley decision plot exhibits the coefficient of friction prediction models via cumulative SHapley Additive exPlanations (SHAP) data. CatBoost has outstanding prediction accuracy, as seen by R2 values ranging from 0.955 to 0.894 for both the training and testing datasets for the COF, as well as 0.992 to 0.885 for surface roughness.