Abstract:

This paper presents the first results of a collaborative experience that is under development as the European ERASMUS+ project SPEET (Student Profile for Enhancing Engineering Tutoring). This project emerges from the potential synergy among a) the huge amount of academic data actually existing at the academic departments of faculties and schools, and b) the maturity of data science in order to provide algorithms and tools to analyse and extract information from what is more commonly referred to Big Data. A rich picture can be extracted from this data if conveniently processed. The main purpose of this project is to apply data mining algorithms to process this data in order to extract information about and to identify student profiles. An idea of the student profile we are referring to within the project scope is, for example: Students that finish degree on time, Students that are blocked on a certain set of subjects, Students that leave degree earlier, etc.

The final goal of the project is the development of an IT tool accessible to higher education institutions that will allow to upload and interactively analyse the correlations, relationships, clusters and groupings that can be extracted from the provided academic data. As intermediate side products of the project there are the software algorithms and tools that will finally be integrated into the IT tool. In fact, once integrated they will allow an interactive upload-and-play activity. However, the partnership understands there are always end users that want to work directly with the core algorithms and to use them to integrate them into eventually existing academic tools.
On the basis of the described scenario, this paper presents the pieces of work that may be of interest to anyone that aims for a direct work with their academic data and to develop its own educational data mining tools. Three kind of basic tools are presented along with examples of its applications to academic data provided by different EU higher education institutions.

First of all academic data is conveniently divided into categorical and performance data of the student as it progresses on the semesters of the degree the student is enrolled on. The main idea is to be able to predict student information as soon as possible by joining the categorical data (static) and the semesters performance (dynamic).

More precisely:
* Performance analysis algorithms: student performance analysis on the basis of categorical and/or performance data; performance for upcoming semesters on the basis of initial information; for explanatory analysis, etc
* Drop-out prediction algorithms: drop-out prediction on the basis of selected categorical information and first semester grades.
* Visualization tools: for visual inspection of the pre-existing data relationships. Dimensional reduction and histogram techniques are applied to project the data on appropriate dimensions suitable for analisys.

The provided tools will be available on open access software repositories and the corresponding description manuals accessible at the project website. Even the main contribution of this work is, obviously, the presentation of the described tools, the topic of educational data mining is of recognized increasing interest among education managers. Therefore, the presentation and categorization of the information profiles that can be extracted from usual academic data is also considered as a contribution of the paper.

Keywords:

International projects, New Experiences for the International Cooperation, Educational Data Mining.