Resumen de A Tool for Automatic Generation of Data Mining Practice Environments

Fernando Terroso Sáenz , Adrián Hernández

• Data analysis subjects are an important part of the computer science degree. In these subjects, seminars and practical sessions where students can practice with different data mining algorithms and techniques are a paramount part. This usually involves the installation and deployment of a common development environment for all the students including libraries, databases, datasets and so forth. Due to the heterogeneity of these environments and the characteristics of the undergraduates, such a configuration and deployment might be an important task in the seminars schedule. Consequently, the present work puts forward an open-source tool for the automatic deployment of data-mining environments based on Docker containers. By means of this tool, the teacher can specify the parameters of the environment to deploy. This results in a set of virtual images comprising the whole practice environment that can be easily distributed among students. This way, we meaningfully optimize the aforementioned deployment step and we also guarantee that all students work with the same practice environment. Finally, we have evaluated the performance of the tool and the level of satisfaction of the students with the generated environments showing quite promising results.