Quantifying uncertainty to optimize power generation: Margaux Zaffran wins L'Oréal-UNESCO 2023 prize

The work of Margaux Zaffran, an engineer and doctoral student specializing in statistics, is motivated by the need for accurate forecasts of electricity market prices - the exchange price between producers and suppliers - as well as probabilistic forecasts of these prices. As opposed to a point forecast (e.g. tomorrow's price at 4pm will be 45€/MWh), a probabilistic forecast seeks to capture the uncertainty of the quantity to be forecast (e.g. there is a 90% chance that tomorrow's price at 4pm will be between 40 and 50€/MWh). Being able to forecast these prices accurately and robustly, i.e. on the basis of a probabilistic forecast, would stabilize energy production planning, and therefore reduce the associated carbon emissions.

The main aim of her research is therefore to improve the stability of the electricity grid, in particular by increasing the profitability of storage systems. On the theoretical side, her work aims at better quantifying the uncertainty of machine learning and artificial intelligence models, so as to minimize the risks associated with these algorithms, even when they are "black-boxes", whatever is the application. And in the longer term, to develop new mathematical tools to better model randomness.

While the main practical application of Margaux Zaffran's current research is forecasting electricity market prices, the methods she is developing are not specific to this application. They can be found in medical diagnostics, where quantifying model uncertainty is essential to avoid using a corrupted model. So, beyond their relevance to energy management, the analysis of these procedures can be usefully applied to many other real-world applications.

Indeed, his latest work will be tested, in collaboration with Traumabase, to predict as accurately as possible whether a trauma patient, taken in charge in an ambulance by the emergency services, will suffer hemorrhagic shock or not, once transferred to hospital.

Although Margaux Zaffran's scientific vocation emerged quite early on in her studies, she first developed a passion for astrophysics and particle physics, before abandoning them because she wanted her work to have a more direct and concrete impact. Then, little by little, through the teachings of her professors, the encounters she had from Master 1 to Master 2, internships and theses, her appetite for statistics gradually grew stronger, and today she has no doubts whatsoever.

Margaux Zaffran is well aware that this grant, which will enable her to shine the spotlight on her career, will make her a role model for many young girls, and help to break down certain unconscious biases and self-censorship. While not every young woman's vocation is science, "I'm firmly convinced that the proportion of young women facing barriers, unconscious or otherwise, among those aspiring to go further down this path is considerable. A better representation of women in science can only be beneficial. On an individual level first, by confirming vocations and creating more motivating examples for young women. And on a collective scale: research advances above all in teams. We all stand to gain from a plurality of viewpoints".

And she projects herself fully into the transmission, while questioning the choice of words, messages or ideas: "How can I ensure that the message received by the reader is the one I really want to convey? How can I present or teach in such a way as to capture the audience's attention while being clear and precise? And conversely, if the authors of this article have chosen this formulation, what does it imply?"