Analyses Multivariées, the RSMAS edition
by Jean-Olivier Irisson, on
As part of this PUF sponsored exchange, students at RSMAS can come to OOV to take classes. In the past two years, one class in particular attracted many RSMAS students: a three weeks intensive course about multivariate statistics. So many students were interested in fact that the funds allocated to student exchange would not have sufficed to cover their travel. In addition, there was not enough room at OOV to host UPMC and RSMAS students at the same time. So we chose to bring the class to them!
After a bit of an administrative struggle, the class was set up at RSMAS, thanks to the sustained efforts of Elizabeth Babcock. And the three professors, Sakina-Dorothée Ayata, Stéphane Gasparini and myself, Jean-Olivier Irisson, travelled to give lectures and supervise labs for two weeks, from March 2nd to March 13th. The topics covered what we usually teach in France: an introduction to the statistical programming language R and statistical GUI software, ordination methods (PCA, CA, MCA, MDS, etc.), clustering methods (hierarchical clustering, k-means, LDA, etc.), and some multivariate data series techniques. Equipped with that knowledge, the students then had a few days to work on a small dataset and presented us their results through a video-conference earlier this week.
From our point of view, the class was a success. The room was very full, with PhD students but also post-doc, and even RSMAS faculty attending. Our "students" were more advanced than the Masters student we usually teach the class to and that prompted a deeper description of the various techniques (plus, some students actually finished the labs!). But now, we are waiting for the feedback from the students to know how the class was received!
Multivariate data analysis is a field of statistics that has a long tradition in France. Some of the techniques we teach have been developed by french researchers (correspondence analysis for example), some even by researchers working in Villefranche (turning points methods for the analysis of data series)! The class itself has a long history and has been taught by various professors over the years in Villefranche. We feel it has matured into something "special"; we are very proud to be teaching it and, now, to bring it overseas. Together with Elizabeth Babcock, we will try to continue offering it after the end of the project. Funding for education-only initiatives, such as this, is even harder to come by than research funding but we keep our hopes up for now!
Here are a few pictures of Sakina warning students about the misuses of PCA, the three professors in a typical Miami setting, and Stéphane and Elizabeth with some of the students at the end of the class.