Splinets – Orthogonal Splines for Functional Data Analysis

Department of Clinical Sciences

Cognitive Systems, Department of Applied Mathematics and Computer Science

Department of Statistics

Abstract

This study introduces an efficient workflow for functional data analysis in classification problems, utilizing advanced orthogonal spline bases. The methodology is based on the flexible Splinets package, featuring a novel spline representation designed for enhanced data efficiency. The focus here is to show that the novel features make the package a powerful and efficient tool for advanced functional data analysis. Two main aspects of spline implemented in the package are behind this effectiveness: 1) Utilization of Orthonormal Spline Bases – the workflow incorporates orthonormal spline bases, known as splinets, ensuring a robust foundation for data representation; 2) Consideration of Spline Support Sets – the implemented spline object representation accounts for spline support sets, which refines the accuracy of sparse data representation. Particularly noteworthy are the improvements achieved in scenarios where data sparsity and dimension reduction are critical factors. The computational engine of the package is the dyadic orthonormalization of B-splines that leads the so-called splinets – the efficient orthonormal basis of splines spanned over arbitrarily distributed knots. Importantly, the locality of \(B\)-splines concerning support sets is preserved in the corresponding splinet. This allows for the mathematical elegance of the data representation in an orthogonal basis. However, if one wishes to traditionally use the \(B\)-splines it is equally easy and efficient because all the computational burden is then carried in the background by the splinets. Using the locality of the orthogonal splinet, along with implemented algorithms, the functional data classification workflow is presented in a case study in which the classic Fashion MINST dataset is used. Significant efficiency gains obtained by utilization of the package are highlighted including functional data representation through stable and efficient computations of the functional principal components. Several examples based on classical functional data sets, such as the wine data set, showing the convenience and elegance of working with Splinets are included as well.

Bader, Michael. 2012. Space-Filling Curves: An Introduction with Applications in Scientific Computing. Vol. 9. Springer Science & Business Media.

Basna, Rani, Xijia Liu, Hiba Nassar, Krzysztof Podgórski, and Hanqing Wu. 2025. “Splinets: Functional Data Analysis Using Splines and Orthogonal Spline Bases.” https://CRAN.R-project.org/package=Splinets.

Basna, Rani, Hiba Nassar, and Krzysztof Podgórski. 2022. “Data Driven Orthogonal Basis Selection for Functional Data Analysis.” Journal of Multivariate Analysis 189: 104868.

———. 2024. “Spline-Based Methods for Functional Data on Multivariate Domains.” Journal of Mathematics in Industry 14 (1): 13.

Benoudjit, Nabil, E Cools, Marc Meurens, and Michel Verleysen. 2004. “Chemometric Calibration of Infrared Spectrometers: Selection and Validation of Variables by Non-Linear Models.” Chemometrics and Intelligent Laboratory Systems 70 (1): 47–53.

Boor, Carl de. 1978. “A Practical Guide to Splines.” In Applied Mathematical Sciences.

Cho, Okkyung, and Ming Jun Lai. 2005. “A class of compactly supported orthonormal B-spline wavelets.” Splines and Wavelets, 123–51.

Goodman, Tim NT. 2003. “A Class of Orthogonal Refinable Functions and Wavelets.” Constructive Approximation 19 (4): 525–40.

Liu, Xijia, Hiba Nassar, and Krzysztof Podgórski. 2022. “Dyadic diagonalization of positive definite band matrices and efficient B-spline orthogonalization.” Journal of Computational and Applied Mathematics 414: 114444.

Mason, JC, Giuseppe Rodriguez, and Sebastiano Seatzu. 1993. “Orthogonal splines based on B-splines – with applications to least squares, smoothing and regularisation problems.” Numerical Algorithms 5 (1): 25–40.

Nassar, Hiba, and Krzysztof Podgórski. 2025. “Periodic Splinets.” Communications in Statistics - Simulation and Computation, 1–16.

Nguyen, Tung. 2015. “Construction of Spline Type Orthogonal Scaling Functions and Wavelets.” Honors Project Paper 19. Illinois Wesleyan University.

Perperoglou, Aris, Willi Sauerbrei, Michal Abrahamowicz, and Matthias Schmid. 2019. “A review of spline function procedures in R.” BMC Medical Research Methodology 19 (1): 46.

Petricoin, Emanuel F, Ali M Ardekani, Ben A Hitt, Peter J Levine, Vincent A Fusaro, Seth M Steinberg, Gordon B Mills, et al. 2002. “Use of Proteomic Patterns in Serum to Identify Ovarian Cancer.” Lancet 359 (9306): 572–77.

Podgórski, Krzysztof. 2021. “Splinets – splines through the Taylor expansion, their support sets and orthogonal bases.” https://arxiv.org/abs/2102.00733.

Qin, K. 2000. “General matrix representations for B-splines.” Vis. Comput. 16: 177–86.

Redd, A. 2012. “A comment on the orthogonalization of B-spline basis functions and their derivatives.” Stat. Comput 22: 251–57.

Redd, Andrew. 2022. “Orthogonalsplinebasis: Orthogonal b-Spline Basis Functions.” https://CRAN.R-project.org/package=orthogonalsplinebasis.

Schumaker, Larry. 2007. Spline Functions: Basic Theory. Cambridge University Press.

Valdez, André Calero. 2021. “Gghilbertstrings: A Fast ’Ggplot2’-Based Implementation of Hilbert Curves.” https://CRAN.R-project.org/package=gghilbertstrings.

Xiao, Han, Kashif Rasul, and Roland Vollgraf. 2017. “Fashion-Mnist: A Novel Image Dataset for Benchmarking Machine Learning Algorithms.” https://arxiv.org/abs/1708.07747.

Zhou, L., J. Z. Huang, and R. J. Carroll. 2008. “Joint Modeling of Paired Sparse Functional Data Using Principal Components.” Biometrika 95: 601–19.