The R-Squared and the Adjusted R-Squared correlation functions were used
in this study to check the model adequacy. The coefficient of
Determination (R-Squared) statistically expresses the explanatory power
of the variables in the model. In Table 3.6, the Coefficient of
Determination (R-Squared) of 0.9345 indicated that more than 93% of the
experimental data explained the independent variables and thus described
the model more explicitly. Adjusted R-Squared topples the R-Squared and
adjusts the number of variables included in the response surface
quadratic model, leading to higher accuracy and precision in describing
the model. An Adjusted R-Squared of 0.8755 indicates that an addition of
a new model term (or some more model terms) will improve the model by
about 87% more likely than would be expected by chance. The Predicted
R-Squared of 0.4980 was not close to the Adjusted R-Squared of 0.8755.
The difference was more than 0.2 and this indicated large block effect
in the response surface quadratic model. Adequacy Precision measures the
signal to noise ratio. A ratio greater than 4 is desirable, therefore
from Table 3.6, the ratio of 10.893 indicated an adequate signal and
hence, this quadratic model could be used to navigate design space.
Table 3.7 Model Coefficients