Stability Selection

Abstract

Summary. Estimation of structure, such as in variable selection, graphical modelling or cluster analysis, is notoriously difficult, especially for high dimensional data. We introduce stability selection. It is based on subsampling in combination with (high dimensional) selection algorithms. As such, the method is extremely general and has a very wide range of applicability. Stability selection provides finite sample control for some error rates of false discoveries and hence a transparent principle to choose a proper amount of regularization for structure estimation. Variable selection and structure estimation improve markedly for a range of selection methods if stability selection is applied. We prove for the randomized lasso that stability selection will be variable selection consistent even if the necessary conditions for consistency of the original lasso method are violated. We demonstrate stability selection for variable selection and Gaussian graphical modelling, using real and simulated data.

Penalized classification for optimal statistical selection of markers from high-throughput genotyping: application in sheep breeds

“…The stability selection method (Meinshausen and Bühlmann, 2010) was used to discover the best subset of variables S stable that will have nonzero weight in the model. Let's assume that we have a generic structure estimation algorithm (i.e.…”

Multivariate Pattern Recognition for Diagnosis and Prognosis in Clinical Neuroimaging: State of the Art, Current Challenges and Future Trends

“…This is still an active field of research in machine learning; e.g., stability selection, which refers to the consistency of features when subsampling the feature space (Meinshausen and Bühlmann 2010), including applications to neuroimaging (Langs et al 2011).…”

Joint sparse canonical correlation analysis for detecting differential imaging genetics modules

“…Especially when the correlation is weak between genetic variant and brain activity, it is more important to detect reliable associations while reducing the FDR (Grellmann et al, 2015). Stability selection is a recently proposed strategy that can better control Type-1 error rate (Meinshausen and Bü hlmann, 2010;Wang et al, 2014), hence it is adopted in our proposed method. The simulations in the supplementary data also demonstrate that the stability selection can yield better results than cross validation.…”

“…On the other hand, since the imaging and genetic correlation is quite low (Grellmann et al, 2015) the selected parameters vary a lot during repeated trials. As an alternative, we apply a hybrid of Monte Carlo validation and stability selection (Meinshausen and Bü hlmann, 2010) to select the parameter s and correlated features. Specifically, we perform random sampling from the original dataset without replacement for B times with the same portion of observations, leading to training samples X bk ; Y bk ; b ¼ 1 .…”