The SCENT model has produced a consistently good performance over all of the data
sets presented so far. The SCENT performance has been better than that of the two
original DNTN networks over almost all the tests. In test 3 the Li performance was
excellent and outperformed the SCENT, but the Racz and Klotz performance was
significantly worse than the SCENT model. The SCENT performance has been good
over both uniform data sets, where the Racz and Klotz network performs poorly, and
data sets with ranges of cluster densities and sizes, that cause the Li network to perform
poorly. The consistency of the SCENT performance over all types of data is notable
and shows that the methods used within the model to assess each data set are both
thorough and flexible.