The analysis of the response surface design model is relatively inaccurate in higher dimensions, which leads to the problem of curse of dimensionality. The dimensionality reduction of model is minimizing the number of factors in the model with minimum loss of information in the data which is useful for concentrating on the detailed analysis about the significant factors in the model. In this paper, an attempt is made to obtain the best model by selecting the significant factors in the first order response surface design model in Bayesian approach. The method is illustrated with suitable examples under imposing and not imposing restrictions on the moment matrix of design. A comparison with classical methods is also presented.

dimensionality reduction, first order model, response surface design, Bayesian approach, variable selection.