A probabilistic class-modelling method based on prediction bands for functional spectral data: Methodological approach and application to near-infrared spectroscopy
Class-modelling; Functional data analysis; Bayesian chemometrics; Spectral predictive distribution; Prediction band; Depth statistic; Multivariate data analysis
Abstract :
[en] Class-modelling methods aim to predict the conformity of new unknown samples with a single target class, using statistical decision rules built exclusively with objects of that class. This article introduces a novel class-modelling method for spectral data. The method uses the concept of beta(%)-prediction band for functional data to classify spectra. The band is defined by an upper and a lower limiting spectra which delimit critical trajectories for beta(%) of future spectra of the target class. It is constructed in three main steps: firstly, a naïve bootstrap sample of calibration spectra is projected onto a parsimonious principal component (PC) basis and their scores are estimated. The posterior predictive distribution of the scores on each PC is estimated using a Bayesian zero-mean normal model. This procedure is repeated on naïve bootstrap estimations of the PCs to obtain the predictive distribution of the scores. These enable to account for all modelling uncertainties including the random deviation of scores from their zero-mean on each PC, uncertainty in the variance of scores (eigenvalue) on each PC, and uncertainty in the PC estimations. Secondly, the predicted scores are back-transformed to the original signal scale to obtain the predictive distribution of future spectra. Thirdly, the predicted spectra are ranked to select the beta(%) most central ones as typical set, whose ranges of variation are used to construct the simultaneous limits of the band. Once the band is constructed, reconstructions of future unknown test spectra by bootstrap PC models are projected onto it, and the extent to which they overlap with it is used to decide their acceptance or rejection. The statistical properties and classification performances of the proposed prediction band are evaluated on real near-infrared datasets and compared to the well-known soft-independent modelling of class analogy (SIMCA) model. The results of the evaluation provide evidence that the proposed prediction band possesses satisfactory predictive performances. It even outperforms the SIMCA while offering attractive advantages like risk-management and straightforward physical interpretability of outlyingness patterns of tested spectra.
Research center :
CIRM - Centre Interdisciplinaire de Recherche sur le Médicament - ULiège
Disciplines :
Pharmacy, pharmacology & toxicology Mathematics
Author, co-author :
Avohou, Tonakpon Hermane ; Université de Liège - ULiège > Département de pharmacie > Chimie analytique
Sacre, Pierre-Yves ; Université de Liège - ULiège > Département de pharmacie > Chimie analytique
Lebrun, Pierre ; Université de Liège - ULiège > Département de pharmacie > Chimie analytique
Hubert, Philippe ; Université de Liège - ULiège > Département de pharmacie > Chimie analytique
Ziemons, Eric ; Université de Liège - ULiège > Département de pharmacie > Chimie analytique
Language :
English
Title :
A probabilistic class-modelling method based on prediction bands for functional spectral data: Methodological approach and application to near-infrared spectroscopy
Publication date :
01 February 2021
Journal title :
Analytica Chimica Acta
ISSN :
0003-2670
eISSN :
1873-4324
Publisher :
Elsevier, Netherlands
Volume :
1144
Pages :
130-149
Peer reviewed :
Peer Reviewed verified by ORBi
Name of the research project :
Vibra4Fake (convention 7517)
Funders :
DGTRE - Région wallonne. Direction générale des Technologies, de la Recherche et de l'Énergie [BE]
Forina, M., Oliveri, P., Lanteri, S., Casale, M., Class-modeling techniques, classic and new, for old and new problems (2008) Chemometr. Intell. Lab. Syst., 93, pp. 132-148
Oliveri, P., Class-modelling in food analytical chemistry: development, sampling, optimization and validation issues – a tutorial (2017) Anal. Chim. Acta, 982, pp. 9-19
Bevilacqua, M., Bucci, R., Magrì, A.D., Magrì, A.L., Nescatelli, R., Marini, F., Chapter 5 - classification and class-modelling (2013) Chemometrics in Food Chemistry, Data Handling in Science and Technology, 28, pp. 171-233. , F. Marini Elsevier Oxford
Pomerantseva, A.L., Rodionova, O.Y., Concept and role of extreme objects in PCA/SIMCA (2014) J. Chemom., 28, pp. 429-438
Rodionova, O.Y., Titova, A.V., Pomerantsev, A.L., Discriminant analysis is an inappropriate method of authentication (2016) Trac. Trends Anal. Chem., 78, pp. 17-22
Pomerantsev, A.L., Acceptance areas for multivariate classification derived by projection methods (2008) J. Chemom., 22, pp. 601-609
Pomerantseva, A.L., Rodionova, O.Y., On the type II error in SIMCA method (2014) J. Chemom., 28, pp. 518-522
Ye Rodionova, O., Oliveri, P., Pomerantsev, A., Rigorous and compliant approaches to one-class classification (2016) Chemometr. Intell. Lab. Syst., 159, pp. 89-96
Ferraty, F., Vieu, P., Nonparametric Functional Data Analysis, Theory and Practice (2006), Springer-Verlag New York
Ferraty, F., Goia, A., Vieu, P., Nonparametric functional methods, New tools for chemometric analysis (2007) Statistical Methods for Biostatistics and Related Fields, pp. 245-264. , W. Härdle Y. Mori P. Vieu Springer-Verlag Berlin
Saeys, W., de Ketelaerea, B., Dariusa, P., Potential applications of functional data analysis in chemometrics (2008) J. Chemometr., 22, pp. 335-344
Wold, S., Sjöström, M., Chapter 12, SIMCA: a method for analyzing chemical data in terms of similarity and analogy (1977) Chemometrics, Theory and Application, 52, pp. 243-282. , B.R. Kowalski American Chemical Society Washington, DC
Vanden Branden, K., Hubert, M., Robust classification in high dimensions based on the SIMCA method (2005) Chemometr. Intell. Lab. Syst., 79, pp. 10-21
Derde, M.P., Massart, D.L., A disjoint modelling technique for pattern recognition based on normal distribution (1986) Anal. Chim. Acta, 184, pp. 33-51
López-Pintado, S., Romo, J., On the Concept of depth for functional data (2009) J. Am. Stat. Assoc., 104, pp. 718-734
Hahn, G.J., Meeker, W.Q., Statistical Intervals: A Guide for Practitioners (1991), p. 392p. , John Wiley & Sons New-York
Krishnamoorthy, K., Mathew, T., Statistical Tolerance Regions: Theory, Applications, and Computation (2009), John Wiley & Sons, Inc. Hoboken
Rozet, E., Hubert, C., Ceccato, A., Walthère, D., Ziemons, E., Moonen, F., Michail, K., Hubert, P., Using tolerance Intervals in pre-study validation of analytical methods to predict in-study results, the fit-for-future-purpose concept (2007) J. Chromatogr. A, 1158, pp. 26-137
(2017) USP 41-NF 36, General Chapter, 1210, Statistical Tools for Analytical Procedure Validation, , US Pharmacopeial Convention Rockville, MD
Gelman, A., Carlin, J.B., Stern, H.S., Dunson, D.B., Vehtari, A., Rubin, D.B., Bayesian Data Analysis (2014), p. 675p. , Chapman and Hall/CRC Boca Raton
Chen, H., Bakshi, B.R., Goel, P.K., Towards Bayesian chemometrics – a tutorial on some recent advances (2007) Anal. Chem. Acta., 602, pp. 1-16
Goldsmith, J., Greven, S., Crainiceanu, C., Corrected confidence bands for functional data using principal components (2013) Biometrics, 69, pp. 41-51
Xiao, L., Zipunnikov, V., Ruppert, D., Crainiceanu, C., Fast covariance estimation for high-dimensional functional data (2016) Stat. Comput., 26, pp. 409-421
Josse, J., Husson, F., Selecting the number of components in principal component analysis using cross-validation approximations (2012) Comput. Stat. Data Anal., 56, pp. 1869-1879
Gavish, M., Donoho, D.L., The optimal hard threshold for singular values is 4/3 (2014) IEEE Trans. Inf. Theor., 60, pp. 5040-5053
Donoho, D.L., Gavish, M., Code Supplement to the Optimal Hard Threshold for Singular Values Is 4/3 (2014), http://purl.stanford.edu/vg705qn9070, (Accessed 27 September 2019) http://purl.stanford.edu/vg705qn9070
Goldsmith, J., Scheipl, F., Huang, L., Wrobel, J., Gellar, J., Harezlak, J., McLean, M.W., Reiss, P.T., Refund: Regression with Functional Data, R Package Version 0 (2018), pp. 1-17. , https://CRAN.R-project.org/package=refund
R Core Team, R., A Language and Environment for Statistical Computing (2018), https://www.R-project.org/, R Foundation for Statistical Computing Vienna, Austria
Carpenter, B., Gelman, A., Hoffman, M.D., Lee, D., Goodrich, B., Betancourt, M., Brubaker, M., Riddell, A., Stan: a Probabilistic programming language (2017) J. Stat. Software, 76
Sun, Y., Genton, M.G., Nychka, D.W., Exact and fast computation of band depth for large functional datasets: how quickly can one million of curves be ranked (2012) Stat, 1, pp. 68-74
Tarabelloni, N., Arribas-Gil, A., Ieva, F., Paganoni, A.M., Romo, J., roahd, Robust Analysis of High Dimensional Data, R Package Version 1 (2018), https://CRAN.R-project.org/package=roahd
Ciza, P.H., Sacre, P.-Y., Waffo, C., Coïc, L., Avohou, T.H., Mbinze, J.K., Ngono, R., Ziemons, E., Comparing the qualitative performances of handheld NIR and Raman spectrophotometers for the detection of falsified pharmaceutical products (2019) Talanta, 202, pp. 469-478
Savitzky, A., Golay, M.J.E., Smoothing and differentiation of data by simplified least squares procedures (1964) Anal. Chem., 36, pp. 1627-1639
Jackson, J.E., Mudholkar, J.S., Control procedures for residuals associated with principal component analysis (1979) Technometrics, 21, pp. 341-349
Malyjurek, Z., Vitale, R., Walczak, B., Different strategies for class model optimization, A comparative study (2020) Talanta, 215
Kucheryavskiy, S., Mdatools – R package for chemometrics (2020) Chemometr. Intell. Lab. Syst., 198, pp. 1-10
Ghahramani, Z., Probabilistic machine learning and artificial intelligence (2015) Nature, 521, pp. 452-459