Jean-Michel Poggi
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Research Interests
Jean-Michel Poggi has a total of 52 co-authors. Here are the top co-authors by number of shared publications:
- Yannig Goude
OSIRIS EDF R&D Bd Gaspard Monge 91120 Saclay France
5 shared publications - Jairo Cugliari
Université Lyon 2, Laboratoire ERIC
4 shared publications - Bruno Portier
INSA Rouen Normandie, Normandie Univ, LMI UR 3226, F-76000 Rouen, France
3 shared publications - Emma Thulliez
INSA Rouen Normandie, Normandie Univ, LMI UR 3226, F-76000 Rouen, France
3 shared publications - Michel Bobbia
ATMO Normandie, 3 place de la pomme d'or, 76000 Rouen, France
3 shared publications - Hui Yan
EDF Lab, 7 bd Gaspard Monge, 91120 Palaiseau, France
2 shared publications
Latest Publications
- Open Access
- 24 June 2026
- Published by Elsevier in Data in Brief
- Vol. 67, 113025
Multi-source dataset of NO2 and PM from mobile and fixed sensors in Rouen (France)
- 24 June 2026
- Published by Elsevier in Data in Brief
- Vol. 67, 113025
Open AccessMulti-source dataset of NO2 and PM from mobile and fixed sensors in Rouen (France)
- 21 July 2025
- Published by Taylor & Francis
- p. 174-182
Statistics and Machine Learning in Industry: Combining Heterogeneous or Multi-Scale Model Outputs
This paper proposes some applied considerations about the prominent role of data and the need to focus on the data produced by algorithms. Applied statisticians could benefit greatly from experimenting more systematically with some machine learning (ML) methods. Sequential learning proposes to optimally combine, and therefore optimally reject, the data produced by algorithms, whether based on statistical methods or not, or by experts. As a statistician, I would like to highlight this important contribution to the methodology of applied statistics by considering two examples of industrial forecasting. These examples use ML algorithms to combine “second-order” data, i.e. data from model outputs from heterogeneous experts and experts on multi-scale aggregates. We will use the sequential prediction paradigm, which allows multiple models of known or unknown internal structure, based on very different assumptions, to be combined in a unified approach. A very attractive feature is that it does not require any prior knowledge about the internal way used by each expert to generate predictions. The key is not to choose the best model or the best resolution, but to exploit the diversity and optimally organise the cooperation of different models or different resolutions.
- 21 July 2025
- Published by Taylor & Francis
- p. 174-182
Statistics and Machine Learning in Industry: Combining Heterogeneous or Multi-Scale Model Outputs
This paper proposes some applied considerations about the prominent role of data and the need to focus on the data produced by algorithms. Applied statisticians could benefit greatly from experimenting more systematically with some machine learning (ML) methods. Sequential learning proposes to optimally combine, and therefore optimally reject, the data produced by algorithms, whether based on statistical methods or not, or by experts. As a statistician, I would like to highlight this important contribution to the methodology of applied statistics by considering two examples of industrial forecasting. These examples use ML algorithms to combine “second-order” data, i.e. data from model outputs from heterogeneous experts and experts on multi-scale aggregates. We will use the sequential prediction paradigm, which allows multiple models of known or unknown internal structure, based on very different assumptions, to be combined in a unified approach. A very attractive feature is that it does not require any prior knowledge about the internal way used by each expert to generate predictions. The key is not to choose the best model or the best resolution, but to exploit the diversity and optimally organise the cooperation of different models or different resolutions.
- 13 March 2025
- Published by Wiley in Applied Stochastic Models in Business and Industry
- Vol. 41 (2)
Digital Twins: A Revolution in Modeling and Simulation
- 13 March 2025
- Published by Wiley in Applied Stochastic Models in Business and Industry
- Vol. 41 (2)
Digital Twins: A Revolution in Modeling and Simulation
- Open Access
- 10 March 2025
- Published by Wiley in Applied Stochastic Models in Business and Industry
- Vol. 41 (2)
Is There a Future for Stochastic Modeling in Business and Industry in the Era of Machine Learning and Artificial Intelligence?
The paper arises from the experience of Applied Stochastic Models in Business and Industry which has seen, over the years, more and more contributions related to Machine Learning rather than to what was intended as a stochastic model. The very notion of a stochastic model (e.g., a Gaussian process or a Dynamic Linear Model) can be subject to change: What is a Deep Neural Network if not a stochastic model? The paper presents the views, supported by examples, of distinguished researchers in the field of business and industrial statistics. They are discussing not only whether there is a future for traditional stochastic models in the era of Machine Learning and Artificial Intelligence, but also how these fields can interact and gain new life for their development.
- 10 March 2025
- Published by Wiley in Applied Stochastic Models in Business and Industry
- Vol. 41 (2)
Open AccessIs There a Future for Stochastic Modeling in Business and Industry in the Era of Machine Learning and Artificial Intelligence?
The paper arises from the experience of Applied Stochastic Models in Business and Industry which has seen, over the years, more and more contributions related to Machine Learning rather than to what was intended as a stochastic model. The very notion of a stochastic model (e.g., a Gaussian process or a Dynamic Linear Model) can be subject to change: What is a Deep Neural Network if not a stochastic model? The paper presents the views, supported by examples, of distinguished researchers in the field of business and industrial statistics. They are discussing not only whether there is a future for traditional stochastic models in the era of Machine Learning and Artificial Intelligence, but also how these fields can interact and gain new life for their development.
- Open Access
- 1 March 2025
- Published by Elsevier in Data in Brief
- Vol. 59, 111439
A dataset of particulate matters measurements from a network of mobile and fixed sensors and monitoring stations in Rouen (Northwestern France)
- 1 March 2025
- Published by Elsevier in Data in Brief
- Vol. 59, 111439
Open AccessA dataset of particulate matters measurements from a network of mobile and fixed sensors and monitoring stations in Rouen (Northwestern France)
- Open Access
- 24 July 2024
- Published by Elsevier in Data in Brief
- Vol. 55, 110768
Deployment of low-cost air quality sensors in Rouen: A dataset of one year of hourly concentrations of gas pollutants
- 24 July 2024
- Published by Elsevier in Data in Brief
- Vol. 55, 110768
Open AccessDeployment of low-cost air quality sensors in Rouen: A dataset of one year of hourly concentrations of gas pollutants
- 23 March 2022
- Published by The R Foundation
Bodi: Boosting Diversity in Regression Ensembles
- Top Cited PapersOpen Access
A Review of Electric Vehicle Load Open Data and Models
The field of electric vehicle charging load modelling has been growing rapidly in the last decade. In light of the Paris Agreement, it is crucial to keep encouraging better modelling techniques for successful electric vehicle adoption. Additionally, numerous papers highlight the lack of charging station data available in order to build models that are consistent with reality. In this context, the purpose of this article is threefold. First, to provide the reader with an overview of the open datasets available and ready to be used in order to foster reproducible research in the field. Second, to review electric vehicle charging load models with their strengths and weaknesses. Third, to provide suggestions on matching the models reviewed to six datasets found in this research that have not previously been explored in the literature. The open data search covered more than 860 repositories and yielded around 60 datasets that are relevant for modelling electric vehicle charging load. These datasets include information on charging point locations, historical and real-time charging sessions, traffic counts, travel surveys and registered vehicles. The models reviewed range from statistical characterization to stochastic processes and machine learning and the context of their application is assessed.
Top Cited PapersOpen AccessA Review of Electric Vehicle Load Open Data and Models
The field of electric vehicle charging load modelling has been growing rapidly in the last decade. In light of the Paris Agreement, it is crucial to keep encouraging better modelling techniques for successful electric vehicle adoption. Additionally, numerous papers highlight the lack of charging station data available in order to build models that are consistent with reality. In this context, the purpose of this article is threefold. First, to provide the reader with an overview of the open datasets available and ready to be used in order to foster reproducible research in the field. Second, to review electric vehicle charging load models with their strengths and weaknesses. Third, to provide suggestions on matching the models reviewed to six datasets found in this research that have not previously been explored in the literature. The open data search covered more than 860 repositories and yielded around 60 datasets that are relevant for modelling electric vehicle charging load. These datasets include information on charging point locations, historical and real-time charging sessions, traffic counts, travel surveys and registered vehicles. The models reviewed range from statistical characterization to stochastic processes and machine learning and the context of their application is assessed.
- Open Access
Random Forests for Time Series
Random forests were introduced in 2001 by Breiman and have since become a popular learning algorithm, for both regression and classification. However, when dealing with time series, random forests do not integrate the time-dependent structure, implicitly supposing that the observations are independent. We propose some variants of the random forests designed for time series. The idea is to replace the standard bootstrap with a dependent bootstrap (i.e block bootstrap) to subsample time series during the tree construction phase to take time dependence into account. We then present two numerical experiments on electricity load forecasting. The first one, at a disaggregated level, is based on an application to load forecasting of a building and illustrate how the variants may perform. The second one is at a more aggregated level (French national forecasting) but focusing on atypical periods. For both, we explore a heuristic for the choice of the block size, the new parameter.
Open AccessRandom Forests for Time Series
Random forests were introduced in 2001 by Breiman and have since become a popular learning algorithm, for both regression and classification. However, when dealing with time series, random forests do not integrate the time-dependent structure, implicitly supposing that the observations are independent. We propose some variants of the random forests designed for time series. The idea is to replace the standard bootstrap with a dependent bootstrap (i.e block bootstrap) to subsample time series during the tree construction phase to take time dependence into account. We then present two numerical experiments on electricity load forecasting. The first one, at a disaggregated level, is based on an application to load forecasting of a building and illustrate how the variants may perform. The second one is at a more aggregated level (French national forecasting) but focusing on atypical periods. For both, we explore a heuristic for the choice of the block size, the new parameter.
- Open Access
Generating experts by boosting diversity
The practical interest of using ensemble methods has been highlighted inseveral works. Sequential prediction provides a natural framework for adapting ensemblemethods to time series data. Main developments focus on the rules of aggregation of aset of experts and examine how to weight and combine the experts. However, very fewwork exist regarding how to choose/generate the experts to include in a given aggregationprocedure. We use the concept of diversity to propose some strategies to enrich the set oforiginal individual predictors. We show how this method is connected to recent theoreticalwork on boosting. We propose a simulation study to illustrate the interest of our approachin the regression setting. An application on real datasets (electricity consumption andpollution data) shows the potentiality of this method for practical forecasting tasks.
Open AccessGenerating experts by boosting diversity
The practical interest of using ensemble methods has been highlighted inseveral works. Sequential prediction provides a natural framework for adapting ensemblemethods to time series data. Main developments focus on the rules of aggregation of aset of experts and examine how to weight and combine the experts. However, very fewwork exist regarding how to choose/generate the experts to include in a given aggregationprocedure. We use the concept of diversity to propose some strategies to enrich the set oforiginal individual predictors. We show how this method is connected to recent theoreticalwork on boosting. We propose a simulation study to illustrate the interest of our approachin the regression setting. An application on real datasets (electricity consumption andpollution data) shows the potentiality of this method for practical forecasting tasks.