The surface quality of any machined part or a product significantly impacts its properties, such as wear and corrosion resistance, fatigue strength, etc. Many safety-instrumented systems and other critical industrial systems require components with a high surface finish. The surface roughness (Ra) is one of the noteworthy measurement quantities that signifies the finishing quality of the machined parts. In this paper, we first present the specifics and Ra data set of an experimental study carried out on a generally used abrasive machining process, i.e., a surface grinder. This is followed by predictive modeling of Ra using machine learning algorithms. The problem is formulated based on the Design Of Experiments (DOE) for the measurement of Ra of EN 8 steel plates using two cutting fluids, namely traditionally used synthetic cutting fluid and an eco-friendly cashew nut shell liquid. In total, 19 experiments were conducted with five input variables set at two levels each. As a part of developing a predictive model, methods (algorithms), namely Linear Regression, Decision Tree, Random Forest, Epsilon–Support Vector Regression (ε-SVR), and K-Nearest Neighbors Regression, are applied to the experimentally acquired data. Computed metrics such as Mean Square Error (MSE), Mean Absolute Percentage Error (MAPE), and Coefficient of Determination (R2), etc. resulted in ε-SVR better than all the other methods, and hence tuning of its hyperparameters is carried out further. Using this developed SVR model, roughness values are predicted that are depicted using a GUI.