Score inference in LASSO regression

We discuss the score statistic to carry out inference on the regression coefficients in LASSO regression. The proposed approach relies on the Induced Smoothing framework and leads to results exhibiting good performance in differ- ent settings, including the high dimensional one n < p. We focus on interval esti- mation where few proposals have been discussed in literature with unsatisfactory results in some settings. We present results from some simulation experiments and an analysis of the well known prostate cancer dataset.