Interpretable functional principal component analysis
- PMID: 26683051
- DOI: 10.1111/biom.12457
Interpretable functional principal component analysis
Abstract
Functional principal component analysis (FPCA) is a popular approach to explore major sources of variation in a sample of random curves. These major sources of variation are represented by functional principal components (FPCs). The intervals where the values of FPCs are significant are interpreted as where sample curves have major variations. However, these intervals are often hard for naïve users to identify, because of the vague definition of "significant values". In this article, we develop a novel penalty-based method to derive FPCs that are only nonzero precisely in the intervals where the values of FPCs are significant, whence the derived FPCs possess better interpretability than the FPCs derived from existing methods. To compute the proposed FPCs, we devise an efficient algorithm based on projection deflation techniques. We show that the proposed interpretable FPCs are strongly consistent and asymptotically normal under mild conditions. Simulation studies confirm that with a competitive performance in explaining variations of sample curves, the proposed FPCs are more interpretable than the traditional counterparts. This advantage is demonstrated by analyzing two real datasets, namely, electroencephalography data and Canadian weather data.
Keywords: EEG; Functional data analysis; Null region; Penalized B-spline; Projection deflation; Regularization; Sparse PCA.
© 2015, The International Biometric Society.
Similar articles
-
Parametric functional principal component analysis.Biometrics. 2017 Sep;73(3):802-810. doi: 10.1111/biom.12641. Epub 2017 Mar 10. Biometrics. 2017. PMID: 28295173
-
Common functional principal components analysis: a new approach to analyzing human movement data.Hum Mov Sci. 2011 Dec;30(6):1144-66. doi: 10.1016/j.humov.2010.11.005. Epub 2011 May 2. Hum Mov Sci. 2011. PMID: 21543128
-
Exploring functional data analysis and wavelet principal component analysis on ecstasy (MDMA) wastewater data.BMC Med Res Methodol. 2016 Jul 12;16:81. doi: 10.1186/s12874-016-0179-2. BMC Med Res Methodol. 2016. PMID: 27406032 Free PMC article.
-
Principals about principal components in statistical genetics.Brief Bioinform. 2019 Nov 27;20(6):2200-2216. doi: 10.1093/bib/bby081. Brief Bioinform. 2019. PMID: 30219892 Review.
-
Considerations for the use of functional principal components analysis in sports biomechanics: examples from on-water rowing.Sports Biomech. 2019 Jun;18(3):317-341. doi: 10.1080/14763141.2017.1392594. Epub 2017 Nov 15. Sports Biomech. 2019. PMID: 29141500 Review.
Cited by
-
Dynamic prediction with time-dependent marker in survival analysis using supervised functional principal component analysis.Stat Med. 2022 Aug 15;41(18):3547-3560. doi: 10.1002/sim.9433. Epub 2022 May 16. Stat Med. 2022. PMID: 35574725 Free PMC article.
-
Effects of Thickness Fraction Process on Physicochemical Properties, Cooking Qualities, and Sensory Characteristics of Long-Grain Rice Samples.Foods. 2022 Jan 14;11(2):222. doi: 10.3390/foods11020222. Foods. 2022. PMID: 35053955 Free PMC article.
-
Interpretable principal component analysis for multilevel multivariate functional data.Biostatistics. 2023 Apr 14;24(2):227-243. doi: 10.1093/biostatistics/kxab018. Biostatistics. 2023. PMID: 34545394 Free PMC article.
-
Predicting the onset of breast cancer using mammogram imaging data with irregular boundary.Biostatistics. 2023 Apr 14;24(2):358-371. doi: 10.1093/biostatistics/kxab032. Biostatistics. 2023. PMID: 34435196 Free PMC article.
-
Time dynamics of COVID-19.Sci Rep. 2020 Dec 3;10(1):21040. doi: 10.1038/s41598-020-77709-4. Sci Rep. 2020. PMID: 33273598 Free PMC article.
Publication types
MeSH terms
LinkOut - more resources
Full Text Sources