This booklet provides an overview of the algorithms that form the foundations of data science and machine learning. It reminds us that an AI solution is indeed based on many technical and mathematical elements (modelling of a regression or classification problem, defining criteria to be optimised, etc.).
All data scientists must master these tools, as well as the different principles and assumptions guaranteeing their correct use. It also highlights that the development of a complete solution is based
on a number of choices that enable the definition of an original solution: choice of model, choice of regularisation, determination of a potential approximation (Monte Carlo type) of the cost function,
selection of variables… The various digital professions focus on understanding and mastering such tools, in order to ensure that the final functionalities of the algorithm meet the needs of the customer.
Although knowledge of these tools is invaluable, many challenges remain. Above all, these relate to measuring the uncertainty of the predictions provided by AI models. It is still a question of developing
simulation methods in order to facilitate the provision of statistical guarantees, which do not exist for many state-of-the-art solutions.
Finally, it seems important to remember the crucial role of data visualisation, analysis and pre-processing. The “data-centric AI competition” promoted by Andrew Ng has recently highlighted this. Numerous models are available, but their performance can be improved with a deeper understanding of the data sets. This highlights once again the crucial role of data analysts and not just that of the complexity of the models.
Special thanks must go to Sylvain Le Corff, teacher of statistics at Telecom Sud Paris, and to the Data Science team: Aurélien Bénard, Houssam Alrachid, Alaeddine Othmani, Safa Hassine and Idriss Brahimi.
Devoteam is a leading consulting firm focused on digital strategy, tech platforms and cybersecurity. By combining creativity, tech and data insights, we empower our customers to transform their business and unlock the future.
With 25 years’ experience and more than 8,500 employees across Europe, the Middle East and Africa,
Devoteam promotes responsible tech for people and works to create better change.
Creative tech for Better Change