Software reliability prediction using package level modularization metrics

摘要

Reliability ensures architectural strength and error free operations of software systems. Software design and architectural measures have been studied as key indicators of faults in software systems. However, association between quantification of reliability prediction using design-level metrics has not been explored. This paper presents a novel approach of developing reliability metrics and it’s prediction using package level metrics. In particular, relevant fault severity information is empirically experimented with package level metrics in an effort-aware classification and ranking scenario. Results obtained hint significant view to predict the reliability of software systems using architectural level metrics. Therefore, the empirical analysis can guide development process to be design-focused and avoid accumulation of faults in implementation phase.