Modified desirability function for optimization of multiple responses / Nasuhar Abd Aziz, Nurul Awanis AS and Siti Nurliana Noraziman

Harrington’s desirability function approach is frequently used to overcome the problem of optimization of multiple responses simultaneously. However, this method will give a huge impact in the presence of outliers. Hence, it is not reliable to use Harrington’s desirability function method to find the optimum responses in this case because it is not resistant to outliers. As an alternative, Modified Geometric Mean (MGM) approach is proposed to estimate the parameter since this approach is resistant to outliers. Numerical example study is carried out to compare the performance of the proposed method with existing procedures. Based on the value of the overall desirability function,D,MGM is better compared with Harrington’s desirability function as it clearly shows that the value of D is larger and the standard error of the MGM approach is smaller. In overall, it is evident that the MGM approach can be an alternative method in dealing with the presence of outliers