Data science and statistical learning

Common core course of mathematics and signal processing of the first year of the École Centrale Méditerranée.

Acquire the essential knowledge and know-how on data processing and learning methods for engineering sciences (applied physics, quality, consulting, risk management, complex systems engineering, logistics, etc.). A wide range of methods is presented with particular attention paid to their basis, implementation and limitations. These methods are illustrated by examples from various fields of application.

• Data and uncertainty modeling,
• Estimation and learning techniques,
• Decision techniques (probabilistic, Bayesian, artificial neural networks),
• Correlated models (Markov analysis and techniques),
• Large deviations and rare events,
• Unsupervised learning and non-parametric estimation.

• Know how to define, implement and characterize different data processing systems for many fields of activity,
• Master the fundamentals of data analysis from industrial and physical systems or for project management in the broadest sense,
• Acquire critical expertise on the choice and limitations of data processing and learning methods,
• Understand the essential factors involved in complex systems.

• Written exam CC1 : 50%
• Reports CC2 : 50%

• Ph. Réfrégier « Noise theory and application to physics » - Springer 2003.
• G. Saporta « Probabilité Analyse des données et statistique » - Editions Technip 1990.
• P.H. Garthwaite, I.T. Jolliffe and B. Jones « Statistical Inference » - Prentice Hall 1995.
• T.M. Cover and J.A. Thomas « Elements of information theory» - Wiley 2006.
• A. Ruegg « Processus stochastiques - Avec applications aux phénomènes d'attente et de fiabilité » - Presses Polytechniques et universitaires romandes 1989.

• G. Berardi
• J. Fade
• F. Galland
• Ph. Réfrégier