Sequential design of the mixed response model based on predicted variance

Many real-world experimental problems involve both qualitative and quantitative responses. While extensive research has addressed experimental design for systems with purely continuous or categorical responses, sequential design strategies for mixed-response systems remain relatively underexplored. This study introduces a sequential design methodology based on Gaussian processes for mixed-response systems, which integrates predictive uncertainties from both response types to guide the design process and improve model prediction accuracy. Simulation results indicate that the proposed approach outperforms traditional space-filling designs, thereby validating its effectiveness as a sequential design strategy.