Grinding being a finishing process, the quality of the ground surface is one of the most important performance evaluation parameters. Grinding process being highly stochastic in nature, surface finish is affected by many factors and experimental evaluation of each factor is a tedious task. In this study, the in-process signals collected using various sensors attached to a cylindrical grinding machine such as Accelerometer and Power are processed, and their features are correlated with a surface finish parameter. This correlation is modelled using gradient boosting algorithm and surface finish obtained is predicted and validated on an industrial application. © 2020 The Authors.