EcoSta 2017: Start Registration
View Submission - EcoSta2017
A0441
Title: A new understanding of LASSO in the presence of measurement errors Authors:  Chi Tim Ng - Hang Seng University of Hong Kong (Hong Kong)
Youngjo Lee - Seoul National University (Korea, South)
Woojoo Lee - Inha University (Korea, South) [presenting]
Abstract: LASSO has been criticized for selecting too many covariates. We show that the LASSO method achieves an optimal prediction when the covariates are subjected to measurement errors. Up to now, all criticisms assume that the covariates are observed without measurement errors, which is not likely to be true in many practical situations. Under measurement errors, the meaning of relevant covariates can be ambiguous. In such a situation, we illustrate that some covariates without an association with the response can potentially be relevant. Neglecting such covariates results in bias in prediction and misinterpretation. To understand the subset of covariates that should be included, a factor model of the covariates is introduced. Two real data examples are provided to illustrate the behaviors of the LASSO method under measurement errors.