The analysis of variance showed that, amongst the four studied parameters, the percentage of
application of the operator Heuristic Mutation in the algorithm (P ) and the reduction factor
(FR) of the temperature in the SA, indeed influence the quality of the hybrid system. At the
same time, the only parametric interaction that induces a statistically significant change in
the HS quality is that between P and FR (the significance level used was 5%). Table 9 shows
the results for the different factors examined.
The interpretation of the results is that if the p-value is less than 0.05, the nullity hypothesis
(H0) is denied and, therefore, there is an impact of the parameter or parametric
interaction on the final distance (target variable) reached by the HS. To assure the validity of
the application of the ANOVA method, the Shapiro-Wilk test (Royston 1995) was employed
on the ANOVA residuals and, with 95% of confidence, the test pointed out normality of the
residuals (p-value = 0.07).