P-splines with an l1 penalty for repeated measures
arXiv:1707.08933 · doi:10.1214/18-EJS1487
Abstract
P-splines are penalized B-splines, in which finite order differences in coefficients are typically penalized with an norm. P-splines can be used for semiparametric regression and can include random effects to account for within-subject variability. In addition to penalties, -type penalties have been used in nonparametric and semiparametric regression to achieve greater flexibility, such as in locally adaptive regression splines, trend filtering, and the fused lasso additive model. However, there has been less focus on using penalties in P-splines, particularly for estimating conditional means. In this paper, we demonstrate the potential benefits of using an penalty in P-splines with an emphasis on fitting non-smooth functions. We propose an estimation procedure using the alternating direction method of multipliers and cross validation, and provide degrees of freedom and approximate confidence bands based on a ridge approximation to the penalized fit. We also demonstrate potential uses through simulations and an application to electrodermal activity data collected as part of a stress study.
arXiv comment54 pages, 26 figures, 5 tables
References in corpus (12)
- A weakly informative default prior distribution for logistic and other regression models
- Smoothing parameter and model selection for general smooth models
- Spike and slab variable selection: Frequentist and Bayesian strategies
- The composite absolute penalties family for grouped and hierarchical variable selection
- Component selection and smoothing in multivariate nonparametric regression
- High-dimensional additive modeling
- Degrees of freedom in lasso problems
- Adaptive piecewise polynomial estimation via trend filtering
- Fast and Flexible ADMM Algorithms for Trend Filtering
- Effective Degrees of Freedom: A Flawed Metaphor
- The Falling Factorial Basis and Its Statistical Applications
- P-splines with an l1 penalty for repeated measures