Defect Prediction Using Historical Bug Repository Metrics in Legacy Systems

Abstract

Defect prediction is an important software maintenance activity in legacy systems because older applications often contain long development histories, undocumented dependencies, repeated patches, and unresolved architectural weaknesses. Historical bug repositories store useful maintenance signals such as defect frequency, reopening rate, module ownership changes, severity distribution, fix duration, and recurrence patterns, which can be used to identify high-risk software components before failure occurs. Existing software quality practices often depend on manual code review, delayed testing, or developer experience, but these approaches may not fully capture long-term defect behavior in aging systems. This article focuses on the use of historical bug repository metrics for predicting defect-prone modules in legacy software environments. The study discusses how repository-derived indicators can be extracted, cleaned, normalized, and mapped to software modules for risk scoring and predictive classification. The article concludes that historical defect metrics can support proactive maintenance planning, reduce regression failures, improve testing prioritization, and extend the operational reliability of legacy systems.