research
Research interests
Latent variable models, Hidden Markov Models
Non-homogeneous hidden Markov chain avec P. AilliotNon-parametric filtering (ex: talk),
Non-linear state-space models (ex: talk),
Parameter estimation (ex: talk),
Multivariate Markov-switching Autoregressive model (ex: talk),
Mixture models for functionnal data with M. Giacofci and M. Morvan ,
Environmental statistics
Space-time stochastic generators for weather variables (wind, temperature, sea state, etc) with P. Ailliot , J. Bessac (ex: review),Correction of meteorological forecast with stochastic models with A. Cuzol and G. Jouan,
Metocean, stochastic wind-wave models,
Data assimilation and analog data assimilation (ex: talk)
Machine learning and health statistics
Decision trees with multivariate splits with A. Poterie,applied to medical statistics, NIR spectrometry data Decision NASH, arthritis, Cancer, metabolomic, intestinal gut and sport performances
Various
2020- ... Membre du bureau et facilitatrice d' AMIES2020- ... Membre nommée du CNU (section 26)
2020- ... Membre du bureau de la commission recherche de l'université de Rennes 1
2020- ... Responsable de l'équipe de statistique de l' IRMAR
Some collaborators
P. Ailliot (UBO, Brest), A. Cuzol (UBS, Vannes), J. Bessac (Argonne, Chicago), M. Giacofci (UR2, Rennes), P. Tandeo (IMT, Brest)Some research projects
Nonparametric Data Assimilation: a data-driven approach (ECOS Sud)
with J. Ruiz (UMI-IFAECI, Argentina), P. Ailliot (UBO, France), P. Tandeo (IMT, France)Understanding and estimating the space-time evolution of geophysical systems constitute a challenge in geosciences. For an efficient restitution of geophysical fields, classical approaches typically combine a physical model based on fluid dynamics equations and remote sensing data or in situ observations. These approaches are generally referred to as data assimilation methods and stated as inverse problems for dynamical processes. A key feature of data assimilation schemes is that they strongly rely on repeated simulations of an explicitly known dynamical model. This may greatly limit their application range as well as their computational efficiency. As an alternative, the amount of observation and simulation data has grown very quickly in the last decades. The availability of such historical datasets strongly advocate for exploring implicit data-driven schemes to build realistic statistical simulations of the dynamics and replace the dynamical model.
DESIRES (ERANET MED)
This project aims to develop an Internet-based, multi-parametric electronic platform for optimum design of desalination plants, supplied by Renewable Energy Sources (RES). The platform will rely upon1) a solar, wind and wave energy potential database,
2) existing statistical algorithms for processing energy-related data,
3) information regarding the inter-annual water needs, 3) a database with the technical characteristics of desalination plant units and the RES components, and
4) existing algorithms for cost effective design, optimal sizing and location selection of desalination plants. DESIRES web site
SEACS (projet inter labex)
Stochastic modEl-dAta-Coupled representationS for the analysis, simulation and reconstruction of upper ocean dynamics This project aims at exploring novel statistical and stochastic methods to address the emulation, reconstruction and forecast of fine-scale upper ocean dynamics.PhD student: Thi Tuyet Trang Chau
Meteobs-bzh
Weather forecast for Bretagne (France), a historical databases from 2007 to 2019/03, Meteobs-bzhWorkshops co-organisation
Workshop "Machine learning and uncertainties in climate simulations", Brest 2021, webpageWorkshop "Statistical models for weather and climate", Rennes, Nov. 2018 webpage
Workshop "Functional Data Analysis 2018", Rennes, Oct. 24-26, 2018, webpage
Workshop "Data Science & environment", Brest, July 04-07, 2017 webpage
"Workshop on Stochastic Weather Generators" , Vannes, May 17-20, 2016 webpage
"Workshop on Stochastic Weather Generators" , Avignon, September 17-19, 2014 webpage
Colloque "Analyse de données spatio-temporelles en océano-météo", 28-29 novembre 2013, Landéda webpage
Colloque "Analyse de données spatio-temporelles en océano-météo", 03-05 juillet 2013, Ile de Berder webpage
"Workshop on Stochastic Weather Generators" May 29th-June 1st, 2012, Roscoff, webpage
Journées "Méthodes statistiques spatio-temporelles en environnement"
"Spatio-temporal Stochastic Models with Environmental and Marine Applications"
Softwares development
xMSannotator_vamos
a R package for network based metabolomic annotation.
Largely inspired from yufree/xMSannotator.
The code has been developped and validated thanks to the
expertise of Shareen Malik.
Available on request for public research (send an
email to valerie dot monbet at univ-rennes1 dot fr).
NHMSAR
a R package including estimation of non homogeneous Markov
switching models simulation validation by parametric
bootstrap Lasso and Scad penalty for the autoregressive and
precision matrices available on CRAN
Metocean Time Series
a Matlab® toolbox (with M. Prevosto, P. Ailliot, P.F. Marteau) including:estimation of parametrical and non parametrical models for multivariate sea state processes (Hs, Tp, Wind speed, Wind directions) parametrical and non parametrical simulation for multivariate sea state processes estimation of persistence statistics such as duration of storms non parametric filtering a bootstrap validation method for complex models download METIS toolbox download documentation (pdf) To get the toolbox for public research and non commercial use, just send me an email : I'll send it to you!