Spotlight on Andrea Lodi

Mining an ocean of data to create smart decision-making systems

Will we live in a smarter society tomorrow? Andrea Lodi’s work is pointing the way to a world in which organizations will make better decisions by efficiently using big data.

Decision-making is a determining factor in the success of a company—and in the evolution of a career in science. Andrea Lodi certainly weighed all the options when he accepted Polytechnique Montréal’s invitation in 2015 to become the Canada Excellence Research Chair in Data Science for Real-Time Decision-Making. Already renowned as one of the world’s leading lights in a field that could be called “the science of smart decision-making,” he would be responsible for Canada’s leading research chair in big data and operations research.

 My chair’s objective is to convert these data into strategic knowledge in real time, to help organizations take the best and most opportune decisions at every point along their innovation, management, production and marketing chain. 

The Italian researcher left the venerable arcades of the University of Bologna for brand-new premises at the Université de Montréal. He settled in the Aisenstadt Building, where the Group for Research in Decision Analysis (known by its French acronym, GERAD); the Institute for Data Valorization; the Montreal Institute for Learning Algorithms; and the Interuniversity Research Centre on Enterprise Networks, Logistics and Transportation have also set up shop.

“I’m fortunate to be able to work closely with these international-calibre teams,” he says. “Moreover, the concentration of expertise here is attracting excellent students from every continent. So, as a professor, I’m spoiled! I think a big part of the future of big data management is being written right here.”

The enthusiasm is palpable on a daily basis: colleagues overrun Lodi’s offices—including researchers, students and industrial collaborators—and the projects are coming in heavy and fast (not to mention that his espresso machine is working overtime).

The reason is that “big data”—a term for data sets that are so large or complex that traditional data processing applications are inadequate—is like a vast digital oilfield, and right now there’s a rush on. Organizations are the eager prospectors, looking to extract its hidden riches. The challenge for them, however, isn’t finding the raw data, it’s refining them into products with high added value, strategically speaking, that in turn can fuel their growth. 

“Companies today have access to a dizzying array of data, in diverse forms—pictures, videos, text and signals, generated by a multitude of platforms,” Lodi explains. “But, it’s difficult to exploit these data, especially in real time, because of their sheer volume and their lack of homogeneity. This makes them incompatible with traditional statistical processing and decision-making methods.”

He continues: “My chair’s objective is to convert these data into strategic knowledge in real time, to help organizations take the best and most opportune decisions at every point along their innovation, management, production and marketing chain.”

There are two aspects to Lodi’s work: first, develop the mathematical and Information Technology tools needed to extract, from the mass of data, an accurate evaluation of a company’s situation; and, second, perfect real-time decision-making methods based on the strategic knowledge that those tools provide. The solutions developed can be adapted to different platforms, including mobile devices.

“Real time is our biggest challenge,” Lodi notes. “To meet it, we’re developing powerful deep-learning and mathematical-optimization algorithms. These work as a chain, sorting the data, detecting relevant information, analyzing it and establishing the decision to be made. The process operates at an extremely fast pace, far more quickly and efficiently than any human could.”

Using the algorithms, the software applications can not only evaluate a situation at a specific moment, but also forecast the future—for example, predict market behaviours and make automated decisions, which are more cost-effective.

“At first glance, our approach may appear fundamental, very mathematical, but clearly there are concrete applications relating to the specific needs of certain organizations,” Lodi explains.“We’re working with, among others, hospitals, municipalities, administrations, transit authorities, high-tech companies and energy utilities.”

While the chairholder is glad to be helping companies fulfil their missions more efficiently, he is particularly pleased about the impact of his work on health-care management. The Chair’s projects in this field combine personalized medicine and care planning. Some projects underway involve telemedicine—for example, applications that collect information and vital signs from patients with chronic conditions, with real-time assessment of whether they need to be hospitalized.

“We’re moving into a world in which, in every field, human decisions will be supported by systems, making them more reliable, faster and more efficient,” Lodi concludes.