UCPH Statistics Seminar: Fabio Feser

Speaker: Fabio Feser from Imperial College London

Title: High-dimensional sparse-group penalised regression models

Abstract: 

Sparse-group models have found increased use in genetics, as they exploit grouping information found in genetic pathways. A new high-dimensional sparse-group approach is proposed for simultaneous variable and group selection, called Sparse-group SLOPE (SGS). SGS achieves false discovery rate control at both variable and group levels by incorporating the sorted L-One penalized estimation (SLOPE) model into a sparse-group framework and exploiting grouping information. A proximal algorithm is implemented for fitting SGS that works for both Gaussian and Binomial-distributed responses. Penalty sequences specific to SGS were derived and shown to provide FDR control under orthogonal designs. Through the analysis of both synthetic and real datasets, the proposed SGS approach is found to outperform other existing lasso- and SLOPE-based models for bi-level selection and prediction accuracy. In particular, SGS is found to be very effective at predicting cases of colitis. Further, the problem of model selection is investigated, with regard to FDR control through the choice of the tuning parameter.