Themes and applications
Decision sciences & Statistics
Operations Research & Decision Analytics
Professors in Operations Research in the IDS department design optimization models and solving methods for decision problems (prescriptive analytics). A wide variety of applications are considered in business and other fields for strategic, tactical or operational decisions: transportation planning and vehicle routing, production planning, network design, facility location, assortment optimization, assignment problems, portfolio optimization, digital operations... Transportation, energy, telecom, healthcare, manufacturing are typical sectors where optimization methods bring value for making better decisions. The research team focuses on combinatorial optimization problems with binary decisions (select or not). The goal is to model these problems, analyze their complexity, and develop solution methods based on mathematical optimization, or approximation heuristics. Both empirical and theoretical validation is provided. The team has a strong expertise in decomposition methods (Branch-and-Price, Branch-and-Cut, Benders decomposition...), bilevel optimization, efficient heuristic or matheuristic methods, and also studies graph problems, choice models and robust and stochastic optimization to deal with data uncertainty, problems dealing with real data (like road networks) and techniques for handling big data, like machine learning.
Professors: Laurent Alfandari, Ivana Ljubic, Claudia Archetti
Econometrics is concerned with the tasks of developing and applying quantitative or statistical methods to the study and elucidation of economic principles. It combines economic or financial theory with probability, statistics and computer sciences to analyze and test relationships. Theoretical econometrics considers questions about the statistical properties of estimators and tests, while applied econometrics is concerned with the application of econometric methods to assess or evaluate theories.
Econometric applications are wide-ranging in social and management sciences, generally when the data are observational rather than obtained via a controlled experiment.
Professors: Guillaume Chevillon, Jeroen Rombouts
Forecasting is the process of estimation in unknown future situations. Its purpose is to generate information, such as probable values, and a range of possible alternatives, depending on the chosen criteria for evaluation of the accuracy, in a context where the future is, potentially loosely, related to present and past. Forecasting techniques rely on probability, statistics, data sciences and time-series econometrics. The focus is on minimizing risk and uncertainty. Applications are wide ranging in economics and increasingly so in management.
Professors: Guillaume Chevillon, Jeroen Rombouts
Over the past decade, statistics and data analysis have undergone drastic changes with the development of high-dimensional statistical inference. The list of fields producing high-dimensional data includes finance, online advertising, genetics. In many of these applications, data analysis is performed under the property of sparsity, that is, assuming that only few variables in a high-dimensional collection of data are effectively relevant. This approach is common for several problems in data analysis such as high-dimensional linear regression, estimation of high-dimensional low rank matrices as well as network models.
Professor: Olga Klopp, Mohamed Ndaoud
Multivariate data analysis and collection
This research stream concentrates on collecting and analyzing huge masses of data where numerous variables are obtained for each properly chosen individual or statistical unit, e.g. answers of many respondents to a variety of questions on different themes in a survey, or different measurements of a company's performance in a study on the company's financial health. The challenge of disentangling complicated interrelationships among various measures and of interpreting both the analytical and the graphical results makes multivariate data analysis a rewarding activity for the investigator. Within the stream of multivariate data analysis, two main approaches are usually distinguished: exploratory and confirmatory. The exploratory, data-driven approach may help in discovering interesting and relevant correlations. It is appropriate when the objective is to identify the structure of the underlying factors that are responsible for co-variation in data with a focus on the variables (dimensionality reduction procedures), on the statistical units (clustering algorithms) or on both (e.g. tandem analysis). The confirmatory theory-driven approach, instead, adapts better to situations where the researcher has hypotheses to test on the number and nature of factors, the merging of available variables into blocks that are the expression of latent concepts, on their parameters and on a network of causal relationships connecting the different latent concepts in accordance with theoretical models. Tools for the modification, the improvement and the assessment of such models are provided in the light of interplay between theory and empirical data.
Professor: Vincenzo Esposito Vinzi, Olga Klopp, Mohamed Ndaoud
The field of computational statistics concerns the computational realization of a statistical workflow such as inference and prediction. The development of novel and principled algorithms to support statistical routines has been the topic of intensive research in the past few decades. In the current age of big data and high-dimensional models, the need to efficiently perform computation and exploit new computing architectures and hardware becomes an all-important challenge. Specific application areas include state space modelling for financial econometrics and epidemiology, and Bayesian statistics for describing epistemological uncertainty.
Professor: Jeremy Heng
IS/IT and strategy
A central theme in organizations concerns how they align their information systems strategy (including information and communications technology strategy) with the overall business strategy. Good alignment ensures, at least to some extent, that the information systems used support the organizational mission and goals. As a consequence, there is likely to be less conflict within organizations resulting from technology adoption and the success of the business strategy becomes feasible.
Professors: Jan Ondrus, Thomas Huber, Julien Malaurent, Thomas Kude
Processes of IS/IT Change
IS/IT evaluation, quality and security
Ever since the Chaos Report in the late 1990s, which suggested that many if not most information systems fail in practice at least to some extent, efforts have been made to improve the success rate of information systems in organizations. In one study of a telecommunications company carried out by researchers at ESSEC and elsewhere, an information systems failure is seen as causing the demise of the company. Evaluation research at ESSEC goes beyond narrow costs and benefits and into satisfying the needs of all stakeholders. In particular our research has shown the necessity of top management involvement. Through researching into the reasons for success and failure, the success rate of information systems has improved greatly over the last few years. Information systems need to be of excellent quality and have high security levels to be successful. We also develop quantitative approaches for assessing the quality of conceptual models, data, and more generally information systems.
Professors: Isabelle Comyn-Wattiau, Nicolas Prat.
IS/IT engineering and development
The development of information systems is often seen as the "core" of the discipline of Information Systems. The study of methodologies, techniques and tools for developing and maintaining information systems is therefore a central theme of practice as well as research. We have developed a deep understanding of how information is represented inside information systems through different forms: models, programs, and data. This allows us to define reverse engineering and evolution methodologies for information systems.
Professors: Isabelle Comyn-Wattiau, Jan Ondrus, Nicolas Prat, Thomas Kude
IS/IT adoption and diffusion
The study of IS innovation adoption is the investigation of how, why and at what rate the innovative ideas, technologies, and services are adopted through cultures and different demographic groups. The research is carried out at both the individual level (Technology Acceptance Model and Theory of Planned Behavior model) and the organizational level (diffusion of innovation theory).
Professors: Yan Li.
E-business and m-business
This domain of research focuses on the business issues raised by the growth in the use of the World Wide Web and the penetration of mobile phones in today's organizations. Faculty interests currently focus on issues in the travel e-distribution and e-marketing fields, and on the area of mobile payments.
Professors: Jan Ondrus.
Knowledge management and decision support
This stream of research concentrates on systems that support managerial actions and decision-making. We conduct research on knowledge management, including the theory of knowledge management and specific technologies like mobile knowledge management. We have developed and are refining methodologies for designing data warehouses and multidimensional systems (OLAP-OnLine Analytical Processing). We have also applied multi-criteria analysis in the context of technology foresight.
Professors: Isabelle Comyn-Wattiau, Jan Ondrus, Nicolas Prat.