Training Data Scientists: A Few Challenges

Gilbert Saporta (1)

1 CEDRIC - MSDMA - CEDRIC. Méthodes statistiques de data-mining et apprentissage

Linked with the Big Data phenomenon, the need of thousands of data scientists in the next years has qualitative and quantitative impacts on the educational system. Academic curricula should include much more information technologies (parallel and distributed computing). Learning in Big commercial software (SAS, SPSS, etc.) becomes less important compared to free environments (R, Python, ScikitLearn, Spark etc.). A data scientist should have three main skills: statistics, computer science, communication. Other curricula are concerned: e.g. economics (see Hal Varian, Big Data: New tricks for econometrics, Journal of Economic Perspectives, 2014), and masters in official statistics: the syllabus of the European Master in Official Statistics (EMOS) needs to be updated. However initial training by universities will not be enough to provide quickly enough specialists. A large part of the solution has to be found in continuous education (or long life training) of statisticians and computer scientists already employed. At CNAM we have developed a professional certificate in Big Data analytics, for statisticians and engineers. Online and distance education (e.g. MOOC) should be developed and there is a clear need for cooperation and mutualisation of efforts, at least at the european level. Learned societies and federations (FenSTATS, EuADS) could be stakeholders in promoting and labelling european courses.

Mots clés: Big Data; Data Science; Learning; Skills

Domaines: Statistiques [stat]; Informatique [cs]

Gilbert Saporta. Training Data Scientists: A Few Challenges. European Data Science Conference, European Association for Data Science, Nov 2016, Luxembourg, Luxembourg. ⟨hal-03953145⟩