Author: Adam Wierzbicki (Univeristy of Warsaw)

Supervisor: prof. Krzysztof Stencel (University of Warsaw)

Detection of software defects has become one of the major challenges in the field of automated software engineering. Numerous studies have revealed that mining data from repositories could provide a substantial basis for defect prediction. In this thesis I introduce my approach towards this problem relying on the analysis of source code history and machine learning algorithms. I describe in detail the proposed computational procedures and explain their underlying assumptions. Following the theoretical basis, I present the results of performed experiments which serve as an empirical assessment of the effectiveness of my methods.

code history, defect, bug-proneness, prediction, repository, metrics, machine learning

D. Software
D.2. Software Engineering
D.2.8 Metrics
D.2.9 Management