Super duplex stainless steels are alloys that have good corrosion resistance properties and are intended for applications in corrosive environments. Due to their chemical composition and microstructure providing high strength and thermal resistance as well as high ductility, the machinability of these alloys is difficult, resulting in longer production cycles and higher costs in terms of more frequent replacement of tools. In this paper, the machinability of the super duplex EN 1.4410 was investigated under environmentally friendly machining process by using cold compressed air as a coolant. Experimental data were generated using the range of selected input parameters and correspondingly analysed surface roughness as output data. Predictive models were developed in order to make a comparison of their prediction performance. In addition, this paper also describes the methodology for optimised development of a particular predictive model. Finally, comparative analysis of the accuracy of predictive models was performed in order to define which model represents the best fit for the analysed experimental data, and also to show validity of the optimisation process.