Multiple smoothing parameters selection in additive regression quantiles

Research output: Contribution to journalArticle


We propose an iterative algorithm to select the smoothing parameters inadditive quantile regression, wherein the functional forms of the covariate effects areunspecified and expressed via B-spline bases with difference penalties on the splinecoefficients. The proposed algorithm relies on viewing the penalized coeffcients asrandom effects from the symmetric Laplace distribution and it turns out to be veryecient and particularly attractive with multiple smooth terms. Through simulationswe compare our proposal with some alternative approaches, including the traditionalones based on minimization of the Schwarz Information Criterion. A real-data analysisis presented to illustrate the method in practice.
Original languageEnglish
Number of pages25
JournalStatistical Modelling
Publication statusPublished - 2020

Fingerprint Dive into the research topics of 'Multiple smoothing parameters selection in additive regression quantiles'. Together they form a unique fingerprint.

  • Cite this