Many real-world optimization problems involve numerical as well as categorical decision variables. Such problems are known as mixed-variable design problems. Evolutionary approaches can naturally work with mixed variables but are computationally expensive. Design space exploration (DSE) allows the designer to understand the interrelation between the decision variables and the nature of the design space. Through DSE, finding regions of interest (ROI) can assist the designer in identifying reduced design space to use evolutionary algorithms for optimization. Interpretable self-organizing maps (iSOM) is a competitive neural network developed for DSE but currently can handle only continuous variables. Optimal scaling (OS) is capable of converting categorical variables into continuous ones. Current work proposes to combine iSOM with OS to visualize design space with mixed variables. The proposed approach is then demonstrated on benchmark nonlinear mathematical problems and two practical engineering design problems involving mixed design variables. The results suggest that the proposed approach is highly efficient in identifying the RoI and obtaining the optima with fewer function evaluations and better precision.