We revisit the classic situation in functional data analysis in which curves are observed at discrete, possibly sparse and irregular, arguments with observation noise. We focus on the reconstruction of individual curves by prediction intervals and bands. The standard approach consists of two steps: first, one estimates the mean and covariance function of curves and observation noise variance function by, e.g. penalized splines, and second, under Gaussian assumptions, one derives the conditional distribution of a curve given observed data and constructs prediction sets with required properties, usually employing sampling from the predictive distribution. This approach is well established, commonly used and theoretically valid but practically, it surprisingly fails in its key property: prediction sets constructed this way often do not have the required coverage. The actual coverage is lower than the nominal one. We investigate the cause of this issue and propose a computationally feasible remedy that leads to prediction regions with a much better coverage. Our method accounts for the uncertainty of the predictive model by sampling from the approximate distribution of its spline estimators whose covariance is estimated by a novel sandwich estimator. Our approach also applies to the important case of covariate-adjusted models.

Department of Mathematics and Statistics Masaryk University Brno Czechia

Zobrazit více v PubMed

Alwan H., Pruijm M., Ponte B., Ackermann D., Guessous I., Ehret G., Staessen J.A., Asayama K., Vuistiner P., Younes S.E., Paccaud F., Wuerzner G., Pechere-Bertschi A., Mohaupt M., Vogt B., Martin P.-Y., Burnier M., and Bochud M., Epidemiology of masked and white-coat hypertension: The family-based SKIPOGH study, PLoS ONE 9 (2014), p. e92522. PubMed PMC

Degras D.A., Simultaneous confidence bands for nonparametric regression with functional data, Stat. Sin. 21 (2011), pp. 1735–1765.

Goldsmith J., Greven S., and Crainiceanu C., Corrected confidence bands for functional data using principal components, Biometrics 69 (2013), pp. 41–51. PubMed PMC

Hall P., Müller H.-G., and Wang J.-L., Properties of principal component methods for functional and longitudinal data analysis, Ann. Stat. 34 (2006), pp. 1493–1517.

Hsing T. and Eubank R., Theoretical Foundations of Functional Data Analysis, with An Introduction to Linear Operators, Wiley, 2015.

Kokoszka P. and Reimherr M., Introduction to Functional Data Analysis, Chapman and Hall/CRC, New York, 2017.

Kraus D., Components and completion of partially observed functional data, J. R. Stat. Soc. Ser. B: Stat. Methodol. 77 (2015), pp. 777–801.

Kraus D., Inferential procedures for partially observed functional data, J. Multivar. Anal. 173 (2019), pp. 583–603.

Liang K.-Y. and Zeger S.L., Longitudinal data analysis using generalized linear models, Biometrika 73 (1986), pp. 13–22.

Liebl D., Inference for sparse and dense functional data with covariate adjustments, J. Multivar. Anal. 170 (2019), pp. 315–335.

Narisetty N.N. and Nair V.N., Extremal depth for functional data and applications, J. Am. Stat. Assoc. 111 (2016), pp. 1705–1714.

Ramsay J.O. and Silverman B.W., Functional Data Analysis, 2nd ed., Springer Series in Statistics, Springer, New York, 2005.

Staniswalis J.G. and Lee J.J., Nonparametric regression analysis of longitudinal data, J. Am. Stat. Assoc. 93 (1998), pp. 1403–1418.

Wood S.N., Generalized Additive Models: An Introduction with R, 2nd ed., Taylor & Francis Group, New York, 2017.

Xiao L., Asymptotic properties of penalized splines for functional data, Bernoulli 26 (2020), pp. 2847–2875.

Yao F., Müller H.-G., and Wang J.-L., Functional data analysis for sparse longitudinal data, J. Am. Stat. Assoc. 100 (2005), pp. 577–590.

Zeileis A., Köll S., and Graham N., Various versatile variances: An object-oriented implementation of clustered covariances in R, J. Stat. Softw. 95 (2020), pp. 1–36.

Zhang X. and Wang J.-L., From sparse to dense functional data and beyond, Ann. Stat. 44 (2016), pp. 2281–2321.