×

zbMATH — the first resource for mathematics

Penalized complexity priors for degrees of freedom in Bayesian P-splines. (English) Zbl 07289471
Summary: Bayesian penalized splines (P-splines) assume an intrinsic Gaussian Markov random field prior on the spline coefficients, conditional on a precision hyper-parameter \(\tau\). Prior elicitation of \(\tau\) is difficult. To overcome this issue, we aim to building priors on an interpretable property of the model, indicating the complexity of the smooth function to be estimated. Following this idea, we propose penalized complexity (PC) priors for the number of effective degrees of freedom. We present the general ideas behind the construction of these new PC priors, describe their properties and show how to implement them in P-splines for Gaussian data.

MSC:
62 Statistics
PDF BibTeX XML Cite
Full Text: DOI
References:
[1] Bartoli, M, Racchetti, E, Delconte, CA, Sacchi, E, Soana, E, Laini, A, Longhi, D, Viaroli, P (2012) Nitrogen balance and fate in a heavily impacted watershed (Oglio River, Northern Italy): In quest of the missing sources and sinks. Biogeosciences, 9, 361-373. doi: 10.5194/bg-9-361-2012. URL http://www.biogeosciences.net/9/361/2012/ (last accessed 28 July 2016).
[2] Belitz, C, Brezger, A, Kneib, T, Lang, S, Umlauf, N (2000) BayesX software for Bayesian inference in structured additive regression models (Technical report). URL http://www.statistik.lmu.de/ bayesx/manual/reference_manual.pdf (last accessed 29 July 2016).
[3] Currie, I, Durbán, M, Eilers, P (2006) Generalized linear array models with applications to multidimensional smoothing. Journal of Royal Statistical Society, Series B 68, 259-280. · Zbl 1110.62090
[4] Delconte, C, Sacchi, E, Racchetti, E, Bartoli, M, Mas-Pla, J, Re, V (2014) Nitrogen inputs to a river course in a heavily impacted watershed: A combined hydrochemical and isotopic evaluation (Oglio River Basin, N Italy). Science of The Total Environment, 466-467, 924-938. doi: http://dx.doi.org/10.1016/j.scitotenv.2013.07.092. URL http://www.sciencedirect.com/science/article/pii/S0048969713008747 (last accessed 28 July 2016).
[5] Eilers, P, Currie, I, Durbán, M (2006) Fast and compact smoothing on large multidimensional grids. Computational Statistics & Data Analysis, 5, 61-76. · Zbl 1429.62020
[6] Eilers, P, Marx, B (1996) Flexible smoothing with B-splines and penalties. Statistical Science, 11, 89-121. · Zbl 0955.62562
[7] Eilers, P, Marx, B (2010) Splines, knots, and penalties. Wiley Interdisciplinary Reviews: Computational Statistics, 2, 637-653.
[8] Fahrmeir, L, Kneib, T (2009) Propriety of posteriors in structured additive regression models: Theory and empirical evidence. Journal of Statistical Planning and Inference, 139, 843-859. · Zbl 1156.62029
[9] Fahrmeir, L, Kneib, T, Lang, S (2004) Penalized structured additive regression for space-time data: A Bayesian perspective. Statistica Sinica, 14, 715-745. · Zbl 1073.62025
[10] Fahrmeir, L, Kneib, T, Lang, S, Marx, B (2013) Regression: models, methods and applications. Berlin: Springer-Verlag. · Zbl 1276.62046
[11] Fong, Y, Rue, H, Wakefield, J (2010) Bayesian inference for generalized linear mixed models. Biostatistics, 11, 397-412. doi: http://dx.doi.org/10.1093/biostatistics/kxp053. URL http://biostatistics.oxfordjournals.org/content/11/3/397.abstract (last accessed 28 July 2016). · Zbl 1437.62460
[12] Frühwirth-Schnatter, S, Frühwirth, R (2007) Auxiliary mixture sampling with applications to logistic models. Computational Statistics & Data Analysis, 51, 3509-3528. doi: http://dx.doi.org/10.1016/j.csda.2006.10.006. URL http://www.sciencedirect.com/science/article/pii/S0167947306003720 (last accessed 28 July 2016). · Zbl 1161.62387
[13] Frühwirth-Schnatter, S, Frühwirth, R, Held, L, Rue, H (2008) Improved auxiliary mixture sampling for hierarchical models of nongaussian data. Statistics and Computing, 19, 479-492. doi: 10.1007/s11222-0089109-4. URL http://dx.doi.org/10.1007/s11222-008-9109-4 (last accessed 28 July 2016).
[14] Frühwirth-Schnatter, S, Wagner, H (2010) Stochastic model specification search for Gaussian and partial non-Gaussian state space models. Journal of Econometrics, 154, 85-100. doi: http://dx.doi.org/10.1016/j.jeconom.2009.07.003. URL http://www.sciencedirect.com/science/article/pii/S0304407609001614 (last accessed 28 July 2016). · Zbl 1431.62373
[15] Frühwirth-Schnatter, S, Wagner, H (2011) Bayesian variable selection for random intercept modeling of Gaussian and nonGaussian data. In Bernardo, JM, Bayarri, MJ, Berger, JO, Dawid, AP, Heckerman, D, Smith, AFM, West, M. Bayesian Statistics 9 pages 165-200, Oxford: Oxford University Press.
[16] Gelman, A (2006) Prior distributions for variance parameters in hierarchical models (comment on article by Browne and Draper). Bayesian Analysis, 1, 515-534. doi: 10.1214/06-BA117A. http://dx.doi.org/10.1214/06-BA117A · Zbl 1331.62139
[17] Hastie, T, Tibshirani, R (1990) Generalized Additive Models. London: Chapman and Hall. · Zbl 0747.62061
[18] Hastie, T, Tibshirani, R, Friedman, J (2009) The Elements of Statistical Learning (Springer Series in Statistics). New York: Springer-Verlag. · Zbl 1273.62005
[19] Jullion, A, Lambert, P (2007) Robust specification of the roughness penalty prior distribution in spatially adaptive bayesian P-splines models. Computational Statistics & Data Analysis, 51, 2542-2558. doi: http://dx.doi.org/10.1016/j.csda.2006.09.027. URL http://www.sciencedirect.com/science/article/pii/S0167947306003549 (last accessed 28 July 2016). · Zbl 1161.62340
[20] Klein, N (2015) sdPrior: Scale-Dependent Hyperpriors in Structured Additive Distributional Regression. R package version 0.3. URL http://CRAN.R-project.org/package=sdPrior (last accessed 28 July 2016).
[21] Klein, N, Kneib, T (2015) Scale-dependent priors for variance parameters in structured additive distributional regression. Bayesian Analysis. doi: 10.1214/15-BA983. URL http://projecteuclid.org/euclid.ba/1448323525 · Zbl 1357.62115
[22] Knorr-Held, L, Rue, H (2002) On block updating in Markov random field models for diasease mapping. Scandinavian Journal of Statistics, 29, 597-614. · Zbl 1039.62092
[23] Kullback, S, Leibler, RA (1951) On information and sufficiency. The Annals of Mathematical Statistics, 22, 79-86. · Zbl 0042.38403
[24] Lang, S, Brezger, A (2004) Bayesian P-splines. Journal of Computational and Graphical Statistics, 13, 183-212. · Zbl 07257866
[25] Rue, H, Held, L (2005) Gaussian Markov Random Fields. London: Chapman and Hall/CRC.
[26] Rue, H, Martino, S, Chopin, N (2009) Approximate bayesian inference for latent Gaussian models using integrated nested laplace approximations (with discussion). Journal of the Royal Statistical Society, Series B, 71, 319-392. · Zbl 1248.62156
[27] Ruppert, D, Wand, P, Carroll, R (2003) Semiparametric Regression (Cambridge Series in Statistical and Probabilistic Mathematics). Cambridge, UK: Cambridge University Press. · Zbl 1038.62042
[28] Simpson, DP, Rue, H, Martins, TG, Riebler, A, Sørbye, SH (2014) Penalising model component complexity: A principled, practical approach to constructing priors. Statistical Science (published before print) URL http://arxiv.org/abs/1403.4630v4”arXiv:1403.4630v4 (last accessed 28 July 2016). · Zbl 1442.62060
[29] Sørbye, SH, Rue, H (2014) Scaling intrinsic Gaussian Markov random field priors in spatial modelling. Spatial Statistics, 8, 39-51. doi: http://dx.doi.org/10.1016/j.spasta.2013.06.004. URL http://www.sciencedirect.com/science/article/pii/S2211675313000407 (last accessed 28 July 2016).
[30] Wakefield, J (2013) Bayesian and Frequentist Regression Methods (Springer Series in Statistics). New York: Springer-Verlag. · Zbl 1281.62014
[31] Wood, S (2006) Generalized Additive Models: An Introduction with R. London: Chapman and Hall/CRC. · Zbl 1087.62082
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. It attempts to reflect the references listed in the original paper as accurately as possible without claiming the completeness or perfect precision of the matching.