
315Model Development and Interpretation
ungraded severities of faults. The completeness value of the DIT is worse with respect to
all the severities of faults. Table7.24shows that most of the studies found that the DIT met-
ric is not related to fault proneness. The class may have less number of ancestors in most of
the studies and is one of the reasons for nonrelation of the DIT metric with fault proneness,
and further investigation is needed. The null hypothesis for the DIT metric is accepted and
the alternative hypothesis is rejected.
Weighted methods per class (WMC) hypothesis is found to be signicant in the LR and
ANN analysis. On the other ...