Bertille Picard

Imagine the following situation: you have a large database with a very large number of characteristics on your individuals. You would like to select the most relevant ones in a data-driven way and you are thinking about implementing a linear regression with variable selection. You hesitate between a LASSO and a Ridge, and you realize that you do not understand the difference between both. This seminar takes you to the frontier of machine learning and bayesian statistics to get a clearer picture - this session requires no previous knowledge in these areas.