Taking adequate managerial decisions in today’s highly complex and interconnected world has become an increasingly complex task. For instance, the 2008 financial crisis has exposed our inability to predict the actual impact of economic decisions on the development of our society. Also, the growing concern (especially regarding agriculture as described in

[1]) and unresolvable uncertainty about climate change critically limits our capacity to foresee under which conditions we will operate in the not-so-distant future and how our current actions might affect these conditions. Indeed, it has become apparent that the most determining decisions that companies, governments and individuals need to make must now necessarily account for the imperfect knowledge of future consequences.

On the flip side, new tools are constantly being developed to help us extract, store and process relevant information from growing databases of historical data. It is actually estimated that the volume of business data worldwide doubles every 1.2 years, with ebay.com accounting currently for an astonishing 50 petabytes (1015 bytes) of such data daily (see [2]). Ironically, one might say that we have never been as much and as little informed in making our decisions as we are today. Indeed, while we can have faith that the new big data scientists will be able to identify unexpected trends in this ocean of information, critical decisions will still need to carefully account for persistent uncertainty if only due to the “historical” (rather than premonitory) nature of this data.

In this context, it is apparent that our society is faced with an information paradox, that is, despite an abundance of information, predicting the consequence of critical decisions is typically impossible. The key challenge in addressing this paradox is ensuring that those who make critical decisions can efficiently process the available information and effectively find the optimal balance between risk and returns for uncertain future consequences. While many important players from both academia and industry have reoriented their efforts to offer services in this regard [3], some significant theoretical and algorithmic developments are still urgently needed to allow decision support tools to evolve from simple analysis software to true decision-making platforms. The Canada Research Chair in decision making under uncertainty was created to answer this urgent and growing need.

Mission of the Research Chair

The mission of the Canada Research Chair in decision making under uncertainty is to develop numerical methods that can either provide decision support or act autonomously in the context of decision-making problems where uncertainty plays a key role. More specifically, the Chair aims to:

  1. Develop and disseminate methods that integrate estimating and optimizing; using such methods, it is possible to extract from available data the information needed for optimal decision-making and to balance between risk and returns in large-scale decision-making problems;
  2. Create at HEC Montréal an excellence and innovation hub in this field.

The Chair’s main objectives include shedding some light on the following questions:

  • How should decisions be made when the exact nature of the factors influencing the performance of the decision is unknown, but a large volume of historical data is available?
  • How can one make optimal decisions that truly reflect the compromises that a decision maker (or a group of decision makers) is willing to make between risk and returns, or between conflicting objectives, while protecting him against inherent cognitive biases?
  • How can optimal decisions be generated autonomously within reasonable computing time when problems take on realistic sizes (i.e. large scale) and need to account for a large (possibly infinite) number of scenarios?

Applications of the research include financial portfolio management, network traffic management, inventory management, energy production and online marketing.

As the recently appointed chairholder, I consider the answers to these questions as cornerstones in allowing decision makers to drastically speed up their reaction time from observations to action while ensuring that these actions reflect their true intent.

The success of this endeavor relies for the most part on two key ingredients: 1) bright and motivated students with a curiosity towards the study and integration of notions from the fields of decision theory, mathematical programming, statistics, artificial intelligence and cognitive science; and 2) industrial partners willing to share this vision and have the patience to support the implementation and validation of innovative methodologies in their field of practice.

 

[1] D. B. Lobell, M. B. Burke, C. Tebaldi, M.D. Mastrandrea, W.P. Falcon, R.L. Naylor, Prioritizing climate change adaptation needs for food security in 2030, Science, Feb. 1st 2008, vol. 319, no. 5863, pages 607-610.

[2] “eBay Study: How to Build Trust and Improve the Shopping Experience”. Knowwpcarey.com. 2012-05-08. Retrieved 2013-03-05.

[3] Recently, many companies such as IBM, Microsoft and Oracle have modified part of their business models to include business analytics software development and consulting services. One can also observe rapid growth in the number of academic programs covering business analytics.