4.4 Checking of Model Adequacy
The residual plots are used to check the goodness of
a fit in regression analysis. Typically, the residual plots are
used to examine if the least squares assumptions are
reasonable to produce unbiased coefficient estimates with
minimum variance. One of the basic assumptions for the
model development is that the errors must be normally
distributed [18]. The normality assumption can be checked
by normal probability plot of residuals. The residuals are the
differences in resilient modulus value between experimental
(observed) and predicted values. Figure 5(a) shows the
normal probability plot of residuals of the values. It is clear
that all the points are quite close to the straight line which
supports that the residuals are normally distributed. The
residuals (errors) and the predicted resilient modulus by the
model are presented in Figure 5(b). It is evident that the
residuals are structure less with no noticeable pattern which
shows that the model is adequate [15].