Estimating a probability that a change will introduce a fault rather
than enhancements or fix is one of the most challenging problems
in the area of software engineering. To approach this task, we
employ two sources of historical data: software repository and bug
tracking system. In my presentation I state motivation of my work,
introduce basic concepts and describe techniques that let us combine
both sources of data. Then I propose an algorithm which decides
whether a change has introduced a defect. We are still working on
employing this information to predict risk of the change.