Machine learning methods and their comparison (using program bugs tracking example)

In this study we discuss Machine Learning methods and compare them based on the program bugs tracking example. This topic is relevant because in the process of developing modern software special means are often needed , which gives the programmers ability to control developing process, receive feedback from the customers and maintain contact with other members of group, especially in big and divided projects. Automatization of given process is necessary, because it will make sure that the majority of tasks are done by machines, rather than humans. Machine Learning Methods are efficient in solving this problem. In this work the module is created, which enables the automatization of assignment the responsible person, whose responsibility is the correction of errors made during the exploitation of the software. Methods are broadly discussed and then the experiments are carried out on the different methods of machine learning. We discuss so-called reports about exceptional situations and automatic assignment of a developer to it in order to correct all of the errors. Also different classifiers are examined. Lastly, the results are compared and analyzed and based on final results the best model is chosen.

Lecture files: