Automated detection of performance regressions in web applications using association rule mining

摘要: Performance testing is an important stage of developing web applications intended to operate with high availability under severe load. However, this process still remains a large extent elaborate, expensive and unreliable. Most often the performance activities are being done manually, significantly affects development time costs. This thesis report describes approach aimed at automating analysis tests by maintaining repository results previously completed test runs comparing them new reveal deviations in software behaviour. Detection degradations executed fast way using well-known data mining techniques. The conducted case studies clearly indicate that suggested may successfully assist engineers detecting regressions evolving software.