Modélisation en Grande Dimension

In this course, we will look at the issue of high dimensionality when the number of covariates (or explanatory variables) exceeds the number of observations. 

We will show the limitations of standard procedures in this context. We present variable selection methods, highlighting their advantages and disadvantages from both a theoretical and a practical point of view. We will present regularization methods adapted to different problems. Finally, we will introduce screening methods to handle the case of ultra-high dimensions when regularization methods are insufficient. There will be practical exercises in R to put into practice the various concepts covered in the course.