Método Ridge em modelo de superficie de resposta: otimização de condições experimentais na elaboração de doce de goiaba

Abstract

The response surface methodology is usually applied in industrial experiments to find optimum conditions for design factors. The purpose of Ridge method is the obtaining of better results for the regression analysis where the multicollinearity problem in a sample is present, severally affecting the estimates obtained by the usual least squares method. This method supplies estimates of the vector of parameters with smaller length than those obtained by the least squares method. This paper aims to show an application for the Ridge method. We used the results of a water activity analysis obtained in an optimization experiment of preserve of guava in mass. The experimental points were analyzed by the response surface methodology (RMS) according to Box & Draper (1997). The work showed that the Rigde method has a good application when the multicollinearity problem exists. The largest addition of sugar caused a water activity reduction of the preserve and this helped in the maintenance of the quality microbiological of the product.