Abstract
Due to the increased importance of machine learning in software and security engineering, effective trainings are needed that allow software engineers to learn the required basic knowledge to understand and successfully apply prediction models fast. In this paper, we present a two-days seminar to teach machine learning-based prediction in software engineering and the evaluation of its learning effects based on Bloom's taxonomy. As a teaching scenario for the practical part, we used a paper reporting a research study on the application of machine learning techniques to predict vulnerabilities in the code. The results of the evaluation showed that the seminar is an appropriate format for teaching predictive modeling to software engineers. The participants were very enthusiastic and self-motivated to learn about the topic and the empirical investigation based on Bloom's taxonomy showed positive learning effects on the knowledge, comprehension, application, analysis, and evaluation level.