In this paper, the limitations of the original proposal are analyzed and a new interpretation methodology is proposed
to guarantee consistent interpretation with future refinements of a current partition. A mixed methodology for cluster
interpretation is proposed that is based on a mixture of interpretation-oriented visualization tools and significance tests.
Class Panel Graphs [11] are used for the visualization, whereas, Test-Value [18] have been imported from the factorial
analysis field of multivariate statistics. The use of Test-Value for cluster interpretation was introduced in [9]. In this paper,
a contribution is introduced on how Test-Value needs to be generalized to better identify the class characteristics. A
modification of the interpretation methodology itself is also introduced to guarantee the consistency of an interpretation
with future refinements in a new nested partition. Sensitivity analysis methods have been used for this purpose