How Machine Learning Has Been Applied In Software Engineering?

Machine Learning (ML) environments are composed of a set of techniques and tools, which can help in solving problems in a diversity of areas, including Software Engineering (SE). However, due to a large number of possible configurations, it is a challenge to select the ML environment to be used for a specific SE domain issue. Helping software engineers choose the most suitable ML environment according to their needs would be very helpful. For instance, it is possible to automate software tests using ML models, where the model learns software behavior and predicts possible problems in the code. In this paper, we present a mapping study that categorizes the ML techniques and tools reported as useful to solve SE domain issues. We found that the most used algorithm is Naive Bayes and that WEKA is the tool most SE researchers use to perform ML experiments related to SE. We also identified that most papers use ML to solve problems related to SE quality. We propose a categorization of the ML techniques and tools that are applied in SE problem solving, linking with the Software Engineering Body of Knowledge (SWEBOK) knowledge areas.