Abstract
Mixtures of regression models (MRMs) are widely used to investigate the relationship between variables coming from several unknown latent homogeneous groups. Usually, the conditional distribution of the response in each mixture component is assumed to be (multivariate) normal (MN-MRM). To robustify the approach with respect to possible elliptical heavy-tailed departures from normality, due to the presence of mild outliers, the multivariate contaminated normal MRM is here introduced. In addition to the parameters of the MN-MRM, each mixture component has a parameter controlling the proportion of outliers and one specifying the degree of contamination with respect to the response variable(s). Crucially, these parameters do not have to be specified a priori, adding flexibility to our approach. Furthermore, once the model is estimated and the observations are assigned to the groups, a finer intra-group classification in typical points and (mild) outliers, can be directly obtained. Identifiability conditions are provided, an expectation-conditional maximization algorithm is outlined for parameter estimation, and various implementation and operational issues are discussed. Properties of the estimators of the regression coefficients are evaluated through Monte Carlo experiments and compared with other procedures. The performance of this novel family of models is also illustrated on artificial and real data, with particular emphasis to the application in allometric studies.
Similar content being viewed by others
References
Aitken AC (1926) A series formula for the roots of algebraic and transcendental equations. Proc R Soc Edinb 45(1):14–22
Aitkin M, Wilson GT (1980) Mixture models, outliers, and the EM algorithm. Technometrics 22(3):325–331
Andrews JL, McNicholas PD (2011) Extending mixtures of multivariate \(t\)-factor analyzers. Stat Comput 21(3):361–373
Andrews JL, McNicholas PD, Subedi S (2011) Model-based classification via mixtures of multivariate \(t\)-distributions. Comput Stat Data Anal 55:520–529
Baek J, McLachlan GJ (2011) Mixtures of common \(t\)-factor analyzers for clustering high-dimensional microarray data. Bioinformatics 27(9):1269–1276
Bagnato L, Punzo A (2013) Finite mixtures of unimodal beta and gamma densities and the \(k\)-bumps algorithm. Comput Stat 28(4):1571–1597
Bagnato L, Punzo A, Zoia MG (2017) The multivariate leptokurtic-normal distribution and its application in model-based clustering. Can J Stat 45(1):95–119
Bai X, Yao W, Boyer JE (2012) Robust fitting of mixture regression models. Comput Stat Data Anal 56(7):2347–2359
Banfield JD, Raftery AE (1993) Model-based Gaussian and non-Gaussian clustering. Biometrics 49(3):803–821
Berkane M, Bentler PM (1988) Estimation of contamination parameters and identification of outliers in multivariate data. Sociol Methods Res 17(1):55–64
Berta P, Ingrassia S, Punzo A, Vittadini G (2016) Multilevel cluster-weighted models for the evaluation of hospitals. METRON 74(3):275–292
Biernacki C, Celeux G, Govaert G (2000) Assessing a mixture model for clustering with the integrated completed likelihood. IEEE Trans Pattern Anal Mach Intell 22(7):719–725
Biernacki C, Celeux G, Govaert G (2003) Choosing starting values for the EM algorithm for getting the highest likelihood in multivariate Gaussian mixture models. Comput Stat Data Anal 41(3–4):561–575
Böhning D (1999) Computer Assisted Analysis of Mixtures and Applications: Meta Analysis, Disease Mapping, and Others, Chapman & Hall/CRC Monographs on Statistics & Applied Probability, vol 81. Taylor & Francis
Böhning D, Dietz E, Schaub R, Schlattmann P, Lindsay B (1994) The distribution of the likelihood ratio for mixtures of densities from the one-parameter exponential family. Ann Inst Stat Math 46(2):373–388
Browne RP, Subedi S, McNicholas PD (2013) Constrained optimization for a subset of the Gaussian parsimonious clustering models. http://arxiv.org/abs/1306.5824
Campbell NA, Mahon RJ (1974) A multivariate study of variation in two species of rock crab of genus Leptograpsus. Aust J Zool 22(3):417–425
Celeux G, Govaert G (1995) Gaussian parsimonious clustering models. Pattern Recogn 28(5):781–793
Celeux G, Hurn M, Robert CP (2000) Computational and inferential difficulties with mixture posterior distributions. J Am Stat Assoc 95(451):957–970
Clarke BR, Davidson T, Hammarstrand R (2017) A comparison of the \(l_2\) minimum distance estimator and the em-algorithm when fitting \(k\)-component univariate normal mixtures. Stat Papers pp 1–20 https://doi.org/10.1007/s00362-016-0747-x
Cuesta-Albertos JA, Gordaliza A, Matrán C (1997) Trimmed \(k\)-means: an attempt to robustify quantizers. Ann Stat 25(2):553–576
Dang UJ, McNicholas PD (2015) Families of parsimonious finite mixtures of regression models. In: Morlini I, Minerva T, Vichi M (eds) Advances in Statistical Models for Data Analysis. Studies in Classification, Data Analysis and Knowledge Organization. Springer, Switzerland pp 73–84
Dang UJ, Browne RP, McNicholas PD (2015) Mixtures of multivariate power exponential distributions. Biometrics 71(4):1081–1089
Dang UJ, Punzo A, McNicholas PD, Ingrassia S, Browne RP (2017) Multivariate response and parsimony for Gaussian cluster-weighted models. J Classif 34(1):4–34
Davies L, Gather U (1993) The identification of multiple outliers. J Am Stat Assoc 88(423):782–792
Dayton CM, Macready GB (1988) Concomitant-variable latent-class models. J Am Stat Assoc 83(401):173–178
Dempster A, Laird N, Rubin D (1977) Maximum likelihood from incomplete data via the EM algorithm. J Roy Stat Soc: Ser B 39(1):1–38
Depraetere N, Vandebroek M (2014) Order selection in finite mixtures of linear regressions. Stat Pap 55(3):871–911
DeSarbo WS, Cron WL (1988) A maximum likelihood methodology for clusterwise linear regression. J Classif 5(2):249–282
Fraley C, Raftery AE, Murphy TB, Scrucca L (2012) mclust version 4 for R: Normal mixture modeling for model-based clustering, classification, and density estimation. Technical report 597, Department of Statistics, University of Washington, Seattle, Washington
Frühwirth-Schnatter S (2006) Finite mixture and Markov switching models. Springer, New York
Galimberti G, Soffritti G (2014) A multivariate linear regression analysis using finite mixtures of \(t\) distributions. Comput Stat Data Anal 71:138–150
García-Escudero LA, Gordaliza A, Mayo-Iscar A, San Martín R (2010) Robust clusterwise linear regression through trimming. Comput Stat Data Anal 54(12):3057–3069
Golam Kibria BM, Safiul Haq M (1999) The multivariate linear model with multivariate \(t\) and intra-class covariance structure. Stat Pap 40(3):263–276
Gómez E, Gómez-Viilegas MA, Marin JM (1998) A multivariate generalization of the power exponential family of distributions. Commun Stat Theory Methods 27(3):589–600
Greselin F, Punzo A (2013) Closed likelihood ratio testing procedures to assess similarity of covariance matrices. Am Stat 67(3):117–128
Grün B, Leisch F (2008) FlexMix version 2: Finite mixtures with concomitant variables and varying and constant parameters. J Stat Softw 28(4):1–35
Hartigan JA (1985) Statistical theory in clustering. J Classif 2(1):63–76
Hastie T, Tibshirani R (1996) Discriminant analysis by Gaussian mixtures. J Roy Stat Soc 58(1):155–176
Hennig C (2000) Identifiablity of models for clusterwise linear regression. J Classif 17(2):273–296
Hennig C (2004) Breakdown points for maximum likelihood estimators of location-scale mixtures. Ann Stat 32(4):1313–1340
Ingrassia S (2004) A likelihood-based constrained algorithm for multivariate normal mixture models. Stat Methods Appl 13(2):151–166
Ingrassia S, Punzo A (2016) Decision boundaries for mixtures of regressions. J Korean Stat Soc 45(2):295–306
Ingrassia S, Rocci R (2007) Constrained monotone em algorithms for finite mixture of multivariate Gaussians. Comput Stat Data Anal 51(11):5339–5351
Ingrassia S, Minotti SC, Punzo A (2014) Model-based clustering via linear cluster-weighted models. Comput Stat Data Anal 71:159–182
Ingrassia S, Punzo A, Vittadini G, Minotti SC (2015) The generalized linear mixed cluster-weighted model. J Classif 32(1):85–113
Jiang W, Tanner MA (1999) Hierarchical mixtures-of-experts for exponential family regression models: approximation and maximum likelihood estimation. Ann Stat 27(3):987–1011
Karlis D, Xekalaki E (2003) Choosing initial values for the EM algorithm for finite mixtures. Comput Stat Data Anal 41(3–4):577–590
Karlsson M, Laitila T (2014) Finite mixture modeling of censored regression models. Stat Pap 55(3):627–642
Klingenberg CP (1996) Multivariate allometry. Advances in Morphometrics. Springer, New York pp 23–49
Knoebel BR, Burkhart HE (1991) A bivariate distribution approach to modeling forest diameter distributions at two points in time. Biometrics 47(1):241–253
Lachos VH, Angolini T, Abanto-Valle CA (2011) On estimation and local influence analysis for measurement errors models under heavy-tailed distributions. Stat Pap 52(3):567–590
Lamont AE, Vermunt JK, Van Horn ML (2016) Regression mixture models: Does modeling the covariance between independent variables and latent classes improve the results? Multivar Behav Res 51(1):35–52
Lange KL, Little RJA, Taylor JMG (1989) Robust statistical modeling using the \(t\) distribution. J Am Stat Assoc 84(408):881–896
Leisch F (2004) FlexMix: A general framework for finite mixture models and latent class regression in R. J Stat Softw 11(8):1–18
Little RJA (1988) Robust estimation of the mean and covariance matrix from data with missing values. Appl Stat 37(1):23–38
Maruotti A, Punzo A (2017) Model-based time-varying clustering of multivariate longitudinal data with covariates and outliers. Comput Stat Data Anal 113(4):475–496
Mazza A, Punzo A, Ingrassia S (2015) flexCWM: Flexible Cluster-Weighted Modeling. http://cran.r-project.org/web/packages/flexCWM/index.html
Mazza A, Punzo A, Ingrassia S (2018) flexCWM. A flexible framework for cluster-weighted models. J Stat Softw pp 1–29
McLachlan G, Krishnan T (2007) The EM algorithm and extensions, Wiley Series in Probability and Statistics, vol 382, 2nd edn. Wiley, New York
McLachlan GJ, Peel D (2000) Finite Mixture Models. Wiley, New York
McNicholas PD (2010) Model-based classification using latent Gaussian mixture models. J Stat Plan Inference 140(5):1175–1181
McNicholas PD, Subedi S (2012) Clustering gene expression time course data using mixtures of multivariate \(t\)-distributions. J Stat Plan Inference 142(5):1114–1127
McNicholas PD, Murphy TB, McDaid AF, Frost D (2010) Serial and parallel implementations of model-based clustering via parsimonious Gaussian mixture models. Comput Stat Data Anal 54(3):711–723
Meng XL, Rubin DB (1993) Maximum likelihood estimation via the ECM algorithm: a general framework. Biometrika 80(2):267–278
Neykov N, Filzmoser P, Dimova R, Neytchev P (2007) Robust fitting of mixtures using the trimmed likelihood estimator. Comput Stat Data Anal 52(1):299–308
Niu X, Li P, Zhang P (2016) Testing homogeneity in a scale mixture of normal distributions. Stat Pap 57(2):499–516
Peel D, McLachlan GJ (2000) Robust mixture modelling using the \(t\) distribution. Stat Comput 10(4):339–348
Punzo A (2014) Flexible mixture modeling with the polynomial Gaussian cluster-weighted model. Stat Model 14(3):257–291
Punzo A, Ingrassia S (2015) Parsimonious generalized linear Gaussian cluster-weighted models. In: Morlini I, Minerva T, Vichi M (eds). Advances in Statistical Models for Data Analysis. Studies in Classification, Data Analysis and Knowledge Organization. Springer International Publishing, Switzerland, pp 201–209
Punzo A, Ingrassia S (2016) Clustering bivariate mixed-type data via the cluster-weighted model. Comput Stat 31(3):989–1013
Punzo A, Maruotti A (2016) Clustering multivariate longitudinal observations: The contaminated Gaussian hidden Markov model. J Comput Gr Stat 25(4):1097–1116
Punzo A, McNicholas PD (2016) Parsimonious mixtures of multivariate contaminated normal distributions. Biometr J 58(6):1506–1537
Punzo A, McNicholas PD (2017) Robust clustering in regression analysis via the contaminated Gaussian cluster-weighted model. J Classif 34(2):249–293
Punzo A, Browne RP, McNicholas PD (2016) Hypothesis testing for mixture model selection. J Stat Comput Simul 86(14):2797–2818
Punzo A, Bagnato L, Maruotti A (2017) Compound unimodal distributions for insurance losses. Insur: Math Econ. https://doi.org/10.1016/j.insmatheco.2017.10.007
Punzo A, Mazza A, McNicholas PD (2018) ContaminatedMixt: An R package for fitting parsimonious mixtures of multivariate contaminated normal distributions. J Stat Softw pp 1–25
Qin LX, Self SG (2006) The clustering of regression models method with applications in gene expression data. Biometrics 62(2):526–533
R Core Team (2013) R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria, http://www.R-project.org/
Ritter G (2015) Robust cluster analysis and variable selection. CRC Press, Baco Raton, CRC Monographs on Statistics & Applied Probability. Chapman & Hall/
Rousseeuw PJ, Driessen KV (1999) A fast algorithm for the minimum covariance determinant estimator. Technometrics 41(3):212–223
Schreuder HT, Hafley WL (1977) A useful bivariate distribution for describing stand structure of tree heights and diameters. Biometrics 33(3):471–478
Schwarz G (1978) Estimating the dimension of a model. Ann Stat 6(2):461–464
Seo B, Kim D (2012) Root selection in normal mixture models. Comput Stat Data Anal 56(8):2454–2470
Skrondal A, Rabe-Hesketh S (2004) Generalized Latent Variable Modeling: Multilevel, Longitudinal, and Structural Equation Models. Interdisciplinary Statistics. Taylor & Francis, Baco Raton
Song W, Yao W, Xing Y (2014) Robust mixture regression model fitting by Laplace distribution. Comput Stat Data Anal 71:128–137
Stephens M (2000) Dealing with label switching in mixture models. J Royal Stat Soc B 62(4):795–809
Subedi S, Punzo A, Ingrassia S, McNicholas PD (2013) Clustering and classification via cluster-weighted factor analyzers. Adv Data Anal Classif 7(1):5–40
Subedi S, Punzo A, Ingrassia S, McNicholas PD (2015) Cluster-weighted \(t\)-factor analyzers for robust model-based clustering and dimension reduction. Stat Methods Appl 24(4):623–649
Tukey JW (1960) A survey of sampling from contaminated distributions. In: Olkin I (ed) Contributions to probability and statistics: essays in honor of Harold Hotelling, Stanford studies in mathematics and statistics. Stanford University Press, California, pp 448–485
Wedel M, Kamakura W (2001) Market segmentation: Conceptual and methodological foundations, 2nd edn. Kluwer Academic Publishers, Boston
Yao W (2012) Model based labeling for mixture models. Stat Comput 22(2):337–347
Yao W, Lindsay BG (2009) Bayesian mixture labeling by highest posterior density. J Am Stat Assoc 104(486):758–767
Yao W, Wei Y, Yu C (2014) Robust mixture regression using the \(t\)-distribution. Comput Stat Data Anal 71:116–127
Acknowledgements
The authors acknowledge the financial support from the grant “Finite mixture and latent variable models for causal inference and analysis of socio-economic data” (FIRB 2012-Futuro in Ricerca) funded by the Italian Government (RBFR12SHVV).
Author information
Authors and Affiliations
Corresponding author
Appendix A: Updates in the first CM-step
Appendix A: Updates in the first CM-step
The estimates of \(\pi _j, {\varvec{B}}_j, \varvec{\varSigma }_j\), and \(\alpha _j, j=1,\ldots ,k\), at the \(\left( r+1\right) \)th first CM-step of the ECM algorithm, require the maximization of
where
Terms which are independent by the parameters of interest have been removed from \(Q_3\). As the three terms on the right-hand side of (14) have zero cross-derivatives, they can be maximized separately.
1.1 A.1 Update of \(\varvec{\pi }\)
The maximum of \(Q_1\left( \varvec{\pi }|\varvec{\vartheta }^{\left( r\right) }\right) \) with respect to \(\varvec{\pi }\), subject to the constraints on those parameters, is obtained by maximizing the augmented function
where \(\lambda \) is a Lagrangian multiplier. Setting the derivative of equation (15) with respect to \(\pi _j\) equal to zero and solving for \(\pi _j\) yields
1.2 A.2 Update of \(\varvec{\alpha }_{{\varvec{Y}}}\)
The updates for \(\varvec{\alpha }\) can be obtained through the first partial derivatives
Equating (16) to zero yields
1.3 A.3 Update of \({\varvec{B}}\) and \(\varvec{\varSigma }_{{\varvec{Y}}}\)
Using properties of trace and transpose, the updates for \({\varvec{B}}\) can be obtained through the first partial derivatives
Equating (17) to the null matrix yields
Finally, the updates for \(\varvec{\varSigma }\) can be obtained through the first partial derivatives
Equating (18) to the null matrix yields
Rights and permissions
About this article
Cite this article
Mazza, A., Punzo, A. Mixtures of multivariate contaminated normal regression models. Stat Papers 61, 787–822 (2020). https://doi.org/10.1007/s00362-017-0964-y
Received:
Revised:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00362-017-0964-y