Mission and research objectives

The HEC Montréal Data Mining Chair is a research chair and the vast majority of its work lies in fundamental research. The breakdown of the Chair’s objectives is as follows:

Breakdown of the Chair’s activities

Quantity and quality of publications                                              60%
Quantity and quality of training activities                                     10%
Supervision of graduate students                                                     15%
Ability to attract students                                                                    5%
Influence and reputation in the academic community                  5%
Awards and distinctions                                                                       5%

The purpose of data mining is to discover, within very large data sets available in both public and private organizations, relevant information that is useful and/or profitable to decision makers.

Although methods have been studied repeatedly over the last decades as part of what was then called “data analysis”, the issues have drastically changed with the huge increase of computing power and the accessibility of very large data sets. Whereas some thirty years ago, researchers were analyzing data sets with 20 to 100 items, it is now common to work with much larger sets containing tens of thousands of items,  sometimes even millions. All the tools available are undergoing a major overhaul to enable them to solve these larger problems. In addition, new methods constantly emerge and applications abound. The data mining sector is of great interest to organizations and attracts some of the best researchers and students.

The HEC Montréal Data Mining Chair has the following goals:

1) Delve into fundamental data mining methods for the three main classes of problems in this area: automatic classification, discrimination and the search for relations.

2) Based on that analysis, develop and/or improve algorithms for specific problems within the above classes.

3) Develop powerful software programs for the resulting algorithms and make them available to both the HEC Montréal community and the broader community of scientific researchers.

4) Apply these methods and use this software to solve concrete problems that occur within organizations and administrations.

To pursue these goals efficiently, the Chair needs to carry out complementary activities both upstream and downstream:

5) On the one hand, the study of fundamental optimization methods that will support the proposed data mining algorithms. This research will focus on exact solution methods as well as heuristic ones, the latter often being the only ones usable with very large data sets.

6) On the other hand, the application of resulting algorithms to the resolution of problems mathematically similar to those of data mining. In this way, the performance of these algorithms can be evaluated and open problems that are themselves of interest can be solved.