Emilie Chautru

Data Science

Course Director

Statistical concepts (population, estimation); dimensionality reduction; best practices (visualization, representativeness, confidentiality, algorithmic bias); supervised learning (empirical risk minimization, regularization, feature selection, and validation).

Elective Course Period (October and January)

Course Director

ATHENS - MP15 - Extreme Value Statistics

Course Director

Geostatistics and Applied Probability (GEOSTAT) Track

Course Director

Regardless of the field of study and the problem at hand, an elective project based on real-world data always involves three aspects: an analysis phase, that is, a critical examination of the available data and an assessment of its suitability for the problem at hand. This simply involves rigorously defining what is being discussed, expressing—if necessary—in scientific terms what is expected from the study, and ensuring that the project has a reasonable chance of success; a modeling phase, because raw data can never be directly manipulated. It is therefore necessary to convert physical measurements into mathematical entities to which the theoretical frameworks covered in the various courses offered at the School can be applied; a synthesis phase, since developing a model is not an end in itself. After mathematical processing, we must therefore find ways to interpret what we have uncovered, even if this means revisiting one or more of the previous steps… In a sense, this first experience in applied geostatistics serves as an introduction to a certain code of ethics regarding the handling of numerical data. In this exercise, the educational effort directed toward external partners who propose research topics is obviously essential. By the end of their third year, students taking this elective have gained an initial overview of issues related to the manipulation of spatial data, and they have grappled in the field with the challenge of reconciling mathematical rigor with the demands of reality. They can now put this introduction to good use, regardless of the career path they choose upon graduating from the School. It is highly likely, in fact, that their professional career will now be in the service of a specific industrial sector that will likely have little connection to what they encountered during their elective work; but because of its fundamental and broad-based nature, the training they have received in this track will certainly prove applicable, even if the term “Geostatistics” is no longer explicitly mentioned! Some representative track topics covered in recent years: numerical models of oil reservoirs; mining estimates and simulations; simulation of flu epidemics; processing of biological or physical data in oceanography and limnology; assimilation of spatiotemporal data in meteorology and climatology; analysis and modeling of pollution data (air, soil, waterways); automatic classification of gemstones; study of the correlation between urban morphology and energy consumption; automatic semantic analysis. Special features of the track: We wish to avoid “monochromatic” cohorts where everyone would focus on petroleum, or mining, or the environment... Furthermore, students are asked to “embrace their specialization” as soon as they choose their track at the end of their first year; it is particularly important that they share their preferences regarding the focus of their specialization project as soon as possible, even if the geostatistics courses themselves have not yet begun: this will enable us to offer them more “tailored” guidance in the direction they have chosen. For students interested in topics related to quantitative finance, the track—even though it does not include courses directly related to the subject—allows them to complete their final thesis in this field. A few supplementary courses that can be taken in this case are listed below.