DEFINING THE “DATA SCIENTIST” PROFESSIONAL PROFILE IN A TRAINING ORGANIZATION: THE CONTRIBUTION OF ARTIFICIAL INTELLIGENCE

The ability to collect, analyze, interpret and use the big amount of data it is strategic for each organizational sector. However, Data Science it is a developing field and new tools and platforms are continuously created or updated. Then, the studies in the scientific literature to define the professional profiles are still ongoing. There is not a unique model, as the data scientist profile is strictly dependent to the organization it belongs to. So, what kind of competencies, skills and knowledge should express a data scientist working in the training sector? What kind of model should we refer to? To answer to this question, we used a Self-Organizing Map, a specific type of neural network. Firstly, we collect the data from the last release of the EDISON Competence Framework documents, considered a benchmark. Then we connected this list of 164 competences, skills and knowledge to the professional profiles identified by the authors: for each of them we assigned a proficiency level. Therefore, we configured the neural architecture, connecting the 164 input neurons with a two-dimensional map and setting the synaptic weights randomly. Finally, we trained the map. The software generated a Kohonen Map which depicts the topological distribution both of the profiles and of the competences of each profile. The training gap between the profiles could be mathematically and economically calculated, as it is a geometrical distance. We believe the HR Department of each organization could use these maps to take decisions about selection, development and training as the map is a support in answering to these core questions:
a) in which area should I place the current or the new human resource to best employ his competences?,
b) how much should I invest in training paths to improve his competences?,
c) how should I assess his performances?.

This is a first step in defining a Data Science Professional Profiles for every type of organizations and it could be used for further research and real case applications.