According to Zhou et al. [19], complexity can be defined with the
Weighted Method Count (WMC) metric1 and the executables
Lines Of Code (LOC) metric. These two metrics are indicators of
problematic classes, which are classes presenting potential
maintenance challenges. Note that WMC and LOC are closely
related; they show a Pearson’s correlation of 94% for this project.
They are still used in conjunction to exclude potentially
descriptive classes with low complexity (WMC) but large number
of LOC.
Extraction of WMC and LOC metrics enables the ordering of each
class of the project from the largest to the smallest. Figure 3 and
Figure 4 show Pre-ID and Post-ID classes, along with the Pareto
distribution for the LOC and the WMC metrics respectively.
Pareto formula is presented in equation (1). The α parameter
defines the shape while the xm parameter defines the scale. The
Pareto distribution parameters closest to the observation
distributions are presented in Table 2.