2. Related Work
The relationships between these rules and the assessing process are evaluated. These
rules are: depth of inheritance tree, fan out, number of children, number of fields, number of
methods, response for class, and weighted methods per class. In [7], statistical analysis
techniques, mainly correlation and logistic regression, are used in order to evaluate the
capability of lack of connection degree rule to predict reliability of object oriented classes
collected from two java Big Data systems. Another class based reliability rules is proposed in
[6]; which detects potential reliability weaknesses of a UML class diagram and points out parts
of the design that need to be improved to reduce the assessing effort. In addition, it proposes a
methodology for improving design reliability. Some studies focus on assessing system reliability
based on object-oriented rules derived from system design [7-9]. In [5] a set of object-oriented
rules are defined to assess the reliability of classes in a Java system.