Robust Algorithms for Change-Point Regressions Using the t-Distribution

Regression models with change-points have been widely applied in various fields. Most methodologies for change-point regressions assume Gaussian errors. For many real data having longer-than-normal tails or atypical observations, the use of normal errors may unduly affect the fit of change-point regression models. This paper proposes two robust algorithms called EMT and FCT for change-point regressions by incorporating the t-distribution with the expectation and maximization algorithm and the fuzzy classification procedure, respectively. For better resistance to high leverage outliers, we introduce a modified version of the proposed method, which fits the t change-point regression model to the data after moderately pruning high leverage points. The selection of the degrees of freedom is discussed. The robustness properties of the proposed methods are also analyzed and validated. Simulation studies show the effectiveness and resistance of the proposed methods against outliers and heavy-tailed distributions. Extensive experiments demonstrate the preference of the t-based approach over normal-based methods for better robustness and computational efficiency. EMT and FCT generally work well, and FCT always performs better for less biased estimates, especially in cases of data contamination. Real examples show the need and the practicability of the proposed method.

[1]  Felipe Osorio,et al.  Detection of a change-point in student-t linear regression models , 2006 .

[2]  Paul Fearnhead,et al.  Changepoint Detection in the Presence of Outliers , 2016, Journal of the American Statistical Association.

[3]  Kaare Brandt Petersen,et al.  On the Slow Convergence of EM and VBEM in Low-Noise Linear Models , 2005, Neural Computation.

[4]  D. Hawkins.  Fitting multiple change-point models to data , 2001 .

[5]  Frederico R. B. Cruz,et al.  Multiple change-point analysis for linear regression models , 2010 .

[6]  A. Munk,et al.  Multiscale change point inference , 2013, 1301.7212.

[7]  Feipeng Zhang,et al.  Robust bent line regression. , 2016, Journal of statistical planning and inference.

[8]  Aliakbar Rasekhi,et al.  Bayesian analysis to detect change-point in two-phase Laplace model , 2016 .

[9]  Fengkai Yang.  Robust Mean Change-Point Detecting through Laplace Linear Regression Using EM Algorithm , 2014, J. Appl. Math..

[10]  V. Muggeo.  Estimating regression models with unknown break‐points , 2003, Statistics in medicine.

[11]  P. Rousseeuw,et al.  Breakdown Points of Affine Equivariant Estimators of Multivariate Location and Covariance Matrices , 1991 .

[12]  Axel Munk,et al.  Heterogeneous change point inference , 2015, 1505.04898.

[13]  Chun Yu,et al.  Robust mixture regression using the t-distribution , 2014, Comput. Stat. Data Anal..

[14]  Gabriela Ciuperca,et al.  Penalized least absolute deviations estimation for nonlinear model with change-points , 2011 .

[15]  Paul Fearnhead,et al.  On optimal multiple changepoint algorithms for large data , 2014, Statistics and Computing.

[16]  T. McMahon.  Using body size to understand the structural design of animals: quadrupedal locomotion. , 1975, Journal of applied physiology.

[17]  M. Muggeo,et al.  segmented: An R package to Fit Regression Models with Broken-Line Relationships , 2008 .

[18]  P. Fearnhead,et al.  Computationally Efficient Changepoint Detection for a Range of Penalties , 2017 .

[19]  Gabriela Ciuperca,et al.  Estimating nonlinear regression with and without change-points by the LAD method , 2011 .

[20]  Gabriela Ciuperca.  A general criterion to determine the number of change-points , 2011 .

[21]  T. Garland.  The relation between maximal running speed and body mass in terrestrial mammals , 2009 .

[22]  Carina Gerstenberger.  Robust Wilcoxon‐Type Estimation of Change‐Point Location Under Short‐Range Dependence , 2018 .

[23]  Paul Fearnhead,et al.  Bayesian detection of abnormal segments in multiple time series , 2014 .

[24]  Hernando Ombao,et al.  FreSpeD: Frequency-Specific Change-Point Detection in Epileptic Seizure Multi-Channel EEG Data , 2018, Journal of the American Statistical Association.

[25]  Rolf Werner,et al.  Study of structural break points in global and hemispheric temperature series by piecewise regression , 2015 .

[26]  Geoffrey J. McLachlan,et al.  Robust mixture modelling using the t distribution , 2000, Stat. Comput..

[27]  G. Willems,et al.  Small sample corrections for LTS and MCD , 2002 .

[28]  Katrien van Driessen,et al.  A Fast Algorithm for the Minimum Covariance Determinant Estimator , 1999, Technometrics.

[29]  Piotr Fryzlewicz,et al.  Wild binary segmentation for multiple change-point detection , 2014, 1411.0858.

[30]  Pulak Ghosh,et al.  Dirichlet Process Hidden Markov Multiple Change-point Model , 2015, 1505.01665.

[31]  Kang-Ping Lu,et al.  Robust algorithms for multiphase regression models , 2020 .

[32]  Anthony C. Atkinson,et al.  The power of monitoring: how to make the most of a contaminated multivariate sample , 2018, Stat. Methods Appl..

[33]  Mohammad Hossein Fazel Zarandi,et al.  A general fuzzy-statistical clustering approach for estimating the time of change in variable sampling control charts , 2010, Inf. Sci..

[34]  Nicolas Vayatis,et al.  A review of change point detection methods , 2018, ArXiv.

[35]  Chuang Wang,et al.  Robust continuous piecewise linear regression model with multiple change points , 2018, The Journal of Supercomputing.

[36]  Miin-Shen Yang.  A survey of fuzzy clustering , 1993 .

[37]  Paul Fearnhead,et al.  A computationally efficient nonparametric approach for changepoint detection , 2016, Statistics and Computing.

[38]  Lixing Zhu,et al.  Heteroscedasticity diagnostics for t linear regression models , 2009 .

[39]  Changliang Zou,et al.  Nonparametric maximum likelihood approach to multiple change-point problems , 2014, 1405.7173.

[40]  S. Robin,et al.  A robust approach for estimating change-points in the mean of an AR(p) process , 2014, 1403.1958.

[41]  Kang-Ping Lu,et al.  A fuzzy classification approach to piecewise regression models , 2018, Appl. Soft Comput..

[42]  Diane J. Cook,et al.  A survey of methods for time series change point detection , 2017, Knowledge and Information Systems.

[43]  P. Rousseeuw.  Least Median of Squares Regression , 1984 .

[44]  Jeremy MG Taylor,et al.  Robust Statistical Modeling Using the t Distribution , 1989 .

[45]  Feng-Chang Xie,et al.  Statistical Diagnostics for Skew-t-Normal Nonlinear Models , 2009, Commun. Stat. Simul. Comput..

[46]  Yong Li,et al.  Bayesian Analysis of Student t Linear Regression with Unknown Change-Point and Application to Stock Data Analysis , 2012 .

[47]  Ping Yang,et al.  Adaptive Change Detection in Heart Rate Trend Monitoring in Anesthetized Children , 2006, IEEE Transactions on Biomedical Engineering.