In the field of teaching technical careers, it is common to find specific topics that students find especially difficult. Regardless of the student’s background, the inherent complexity due to the high degree of abstraction of some of these terms is combined with the challenge of defining them in a precise manner. In many cases, these concepts can have different meanings depending on the context.

Two of these terms, which are fundamental pillars in the development of a technical career, are modelling and model identification. Part of the problem lies in the pre-conceived idea students may have of what a model is, ranging from a person to a model of a building or a reference role to imitate.

The module “Artificial Intelligence Applied to Control” (IACC), which is optional in the degrees of Computer Engineering, Computer Engineering and Software Engineering taught at the Faculty of Computer Science at the Complutense University of Madrid, faces this issue every academic year.

In order to help the students to assimilate the terminology and concepts, a series of practical exercises associated to this topic are proposed. Specifically, and to facilitate their skill in the use of computational tools, data-driven identification of a real system is proposed. First, an example consisting on a time series of data from the field of renewable energy, specifically wind energy, is provided. It has been found that topics related to sustainability and the environment are particularly attractive to students.

Using the Internet, they search for a public, open database from which the students can collect data on the evolution of the wind over the past few years. They must then graph it, carry out a visual analysis, and then apply a tool to obtain a function that represents this sequence as faithfully as possible.

They then choose a system or signal of interest, from variations in car prices, population or temperature trends, cryptocurrency prices, CPU power, etc., to commercial flight trajectories. Using a computational program, they must identify the system, whose input and output data are known. There are some available libraries that help them to try to fit the data to different functions. The students then must choose among the available approaches which on gives better results.

The most relevant part of this practice is the justification of the selection made, as it requires to know how to interpret not only the graph but also the errors, which can be calculated, the complexity of the model, the parameters involved, etc.

The students are also requested to justify how to validate the model, and how to use it in future values prediction. They are encouraged to think of a company/ enterprise that might be interested in their prediction model, and to desing a dissemination plan such that it described the characteristics and potential usefulness so that the enterprise would be interested in acquiring their product (prediction model).

These activities and exercises promote not only the acquisition of new academic concepts but also the development of a series of skills such as:
- Computational skills.
- Critical thinking.
- Interpretation of graphs and results.
- Presentation and communication of a development made by them.

Keywords: Computational tools, learning by doing, motivation, technical degree.