Yannick Léo is Associate Partner & Data Science Director of Emerton Data with over 10 years of experience in the Data and AI domains. Currently, he leads the data science and scientific research at Emerton Data, where he specializes in developing data products and defining data strategies.
Before joining Emerton Data, Yannick worked as a data scientist at Sodexo, where he supported the launch of their data & AI journey, defined data science methodologies, and developed various data products to the highest standards.
He has worked in various environment such as research labs (INRIA), startups (Adeline), large corporations (Sodexo) and data strategy consulting, bringing his expertise to a wide range of industries, including food and services, energy, government, and consumer goods.
With his extensive background in data science, including expertise in statistics, machine learning, and deep learning, Yannick has implemented scientific approaches in various data science domains such as computer vision, natural language processing, optimization, and time series/forecasting.
Yannick holds a Ph.D. in Computer Science from Ecole Normale Supérieure de Lyon, followed by a 1-year post-doctorate at INRIA. He has published several scientific articles in big data, graph theory, and machine learning fields in first-class journals and conferences. Yannick's research has always been multi-disciplinary, combining data and domain expertises to innovate and bring value.
The unique analysis of data from more than 30 billion geolocalized steps, carried out jointly by Emerton Data and WeWard, sheds light on the various mobility phenomena observed since the announcement of confinement in France.
The results presented in this report provide food for thought on the one hand on the implications of a generalized and uniform confinement across the country, and on the other hand on the methods of implementing deconfinement or a revised confinement. The announcement of confinement, made a few days before its effective date, provoked many long-distance pre-confinement trips: around 20% of French people changed departments, mainly in the departments adjoining their place of residence.
To what extent does this massive departure for the 18-45 age group concern family reunification with parents who are vulnerable to covid-19?
This paper introduces X-SHAP, a model-agnostic method that assesses multiplicative contributions of variables for both local and global predictions.
This methodtheoretically and operationally extends the so-called additive SHAP approach. Itproves useful underlying multiplicative interactions of factors, typically arisingin sectors where Generalized Linear Models are traditionally used, such as ininsurance or biology.
We test the method on various datasets and propose a set oftechniques based on individual X-SHAP contributions to build aggregated multiplicative contributions and to capture multiplicative feature importance, that wecompare to traditional techniques.
