Optimizing Supply Chains Through High-Dimensional Statistics

CY Initiative: Could you introduce yourself and tell us more about your career and your primary research themes?

Olga Klopp: I have a PhD in mathematics and a habilitation to direct research (HDR) in high-dimensional statistics. Before joining ESSEC in 2017 as a professor in the Information Systems, Data Analysis and Operations department, I was a lecturer at the University of Nanterre. At ESSEC, I am also director of the double degree programs with École Centrale Supelec, and ENSAE Paris.

Generally speaking, my research focuses on statistics and data analysis, particularly high-dimensional statistics. These statistics extend beyond the classical framework that limits the number of parameters. Modern datasets often contain numerous parameters to estimate, but only a limited number of observations.  

For more than 20 years now, the field of high-dimensional statistics has been developing strongly. For example, high-dimensional data often refers to genetic data. A gene is a very long sequence, generating a huge amount of data. Examining genetic mutations and the relationships between specific gene segments and diseases requires processing vast amounts of data. We can't sequence as many genes as we need to.

CY Initiative: You won the CY Initiative's 2023 call for proposals. What does this project involve, and what is its objective? 
 
Olga Klopp: The BESCHAP project, which I'm leading jointly with my colleague Ivana Ljubic, also a professor at ESSEC, is an applied research project that are specifically aimed at developing new forecasting methods for supply chains. The idea arose from a discussion with a company facing problems in this area. The aim is to develop a new forecasting method based on a specific model known as Network Autoregressive Process (NAR).

By utilizing historical data and values from companies within a supply network, as well as certain observed variables, such as the price of oil, we will develop a forecasting model. In a supply chain, we find a network where companies are connected to generate flows. It is essential to consider these connections and models in our current analysis. The main idea is therefore that this new method based on the NAR model can produce more effective forecasts for supply chains. Moreover, we believe that if this model is sufficiently robust, it can also incorporate critical sustainability concepts, such as the calculation of CO2 emissions, in order to control pollution linked to transport and goods.

Our aim is to enable companies to make more effective forecasts, and to integrate these forecasts with the distribution process, specifically the organization of transportation. Currently, this issue is recognized as significant, but there is very little research work on combining forecasting with transport optimization. These two issues are often studied separately, but we believe that we can combine the two to improve logistics as a whole. Our goal is to develop a new general-purpose forecasting algorithm, applicable not only to supply chains but also to other fields. This algorithm should be able to introduce variables that take account of weather and seasons.

CY Initiative: In concrete terms, how will the project be rolled out? How does the project address current issues? 

Olga Klopp: We have made significant progress on this topic. First, we analyzed a large amount of data, and subsequently developed an algorithm that we will need to test in a real-world case. This is the most complex part, as the data is linked to the health of the company and its sales, which often makes it difficult to retrieve, as it is often confidential.  
We are currently working with a start-up specializing in food chains, Califrais, the official digital & logistics operator at Rungis. They have forecasting challenges, particularly in terms of transport optimization. Their objectives are to save time and reduce CO2 emissions. The need is based on two areas: forecasting and transport optimization.   

This is a major challenge for companies today. Enhanced forecasting allows companies to respond more effectively to market changes. They can avoid overstocking, overproduction, wastage and stock-outs. However, the challenge extends beyond business concerns to societal issues, including waste management, transportation costs, and carbon footprints. 

CY Initiative: Why did you apply for this Call for proposals?  

Olga Klopp: The purpose of this project and this call for proposals is to determine whether this model is effective, if it makes sense, and then to apply for a collaborative ANR with a company. We are already in discussions with a start-up specializing in food chains to work on this next stage.
 

Learn More

• Olga Klopp's Page on the ESSEC Website
• All the CY Initiative Projects