Several years ago we presented an experimental, discovery-learning approach to the active learning of greedy algorithms. This paper presents GreedEx, a visualization tool developed to support this didactic method. The paper states the design goals of GreedEx, makes explicit the major design decisions adopted, and describes its main characteristics in detail. It also describes the experience of use, the usability evaluations conducted, and the evolution of GreedEx in these years in response to the findings of the usability evaluations. Finally, the positive results obtained in an evaluation of educational effectiveness are shown. The paper has three main contributions. First, the GreedEx system itself is an innovative system for experimentation and discovery learning of greedy algorithms. Second, GreedEx is different from other visualization systems in its support to higher levels of learning, in particular evaluation tasks. Finally, GreedEx is an example of a medium-term research project, where an educational system was designed from explicit learning goals and was later refined in a user-centered design process involving instructors and students, before carrying out a successful evaluation of educational effectiveness.

An evaluation of the educational effectiveness of a didactic method for the active learning of greedy algorithms is presented. The didactic method sets students structured-inquiry challenges to be addressed with a specific experimental method, supported by the interactive system GreedEx. This didactic method has been refined over several years of use. Additional elements are lecture contents and the scheduling of classes and lab sessions. Learning gain in the topic of greedy algorithms was measured in the short term for two groups of students: an experimental group taught with the new didactic method, and a control group taught with a traditional approach. The results show a significant learning gain improvement in the experimental group, while the students taught with a traditional method had little learning gain. In addition, the levels of Bloom's taxonomy at which improvements occurred were identified. In the control group, improvements were found at the knowledge level. In the experimental group, however, improvements were found at the knowledge and the comprehension levels, although not at the analysis level. These results are encouraging and indicate directions for future research, as analysis skills are important in algorithm courses.

No abstract