High-dimensional inference: Lasso estimates are either explicitly 0 (if a variable is not selected) or non-zero but biased against 0. How can we construct confidence intervals or compute p-values for these methods? Re-sampling is tricky because the behaviour of the selection is non-smooth: you're either "in" and have an estimated coefficient value or you are "out" and have parameter value 0. In the lecture we will discuss re-sampling and de-biasing methods for lasso methods.

Recommended reading:

HDI paper: High-Dimensional Inference: Confidence Intervals, p-Values and R-Software hdi Ruben Dezeure, Peter B¨uhlmann, Lukas Meier and Nicolai Meinshausen

The python version of the package can be found here: https://github.com/ja-che/hidimstat

Stability selection paper: Journal of the Royal Statistical Society Series B Statistical Methodology - 2010 - Meinshausen - Stability selection.pdf