Nowadays, software development teams follow modern principles regarding the software development life cycle and use many tools, such as version control systems, bug tracking systems etc., in order to improve their productivity. The popularity and the intensive use of such tools and systems, has gathered a big amount of information regarding every stage of the software development process. By utilizing and analyzing these data, we can extract valuable information and build tools that contribute to the field of qualitative software development. New trends around software development processes aim at proper distribution of tasks to the team, flexibility in dynamic situations and development of a timetable that corresponds to reality. The achievement of the above in large open source organizations can be achieved through the analysis of software development methods and the design of systems that automate relevant processes. This diploma proposes an end-to-end system that contributes to the research of bug fix time prediction, by applying information retrieval techniques. More precisely, the designed system collects and analyzes data from GitHub repositories. The system classifies software issues according to their predicted fix time. Our approach is multilevel, taking into consideration the features title, description, assignee and labels of a bug report. A subsystem is designed for each of these features. Subsystems analyze previous data and generate a score that represents the probability of participation for each issue in every class. Finally, classification is performed by a neural network that aggregates every subsystem’s scores. Moreover, data processing techniques are used in order to cope with the particularities shown in the datasets of open source software repositories. The proposed system is trained and evaluated in a dataset that consists of 11099 issues from 26 big Java repositories in GitHub. Experiments show that our system has satisfactory efficiency, especially when it comes to binary classification where high evaluation metrics are observed.