Abstract
The Bayesian literature on the change point problem deals with the inference of a change in the distribution of a set of time-ordered data based on a sample of fixed size. This is the so-called “retrospective or off-line” analysis of the change point problem. A related but different problem is that of the “sequential” change point detection, mainly analyzed from a frequentist viewpoint. While the former typically focuses on the estimation of the position in which the change point occurs, the latter is a testing problem which has a natural formulation as a Bayesian model selection problem. In this paper we provide such a Bayesian formulation, which generalizes previous formulations such as the well-known CUSUM stopping rule. We show that the conventional improper priors (also called non-informative, objective or default), cannot be used either for sequential detection of the change or for retrospective estimation. Then, we propose objective intrinsic prior distributions for the unknown model parameters. The normal and Poisson cases are studied in detail and examples with simulated and real data are provided.
Similar content being viewed by others
References
Abramowitz M, Stegun IA (1970) Handbook of mathematical functions. Dover Publications Inc., New York
Balke NS (1993) Detecting level shifts in time series. J Business Econ Stat 11:81–92
Berger JO (2000) Bayesian analysis: a look at today and thoughts of tomorrow. J Amer Stat Assoc 95:1269–1276
Berger JO, Bernardo JM (1992) On the development of the reference prior method. In: Bernardo JM et al. (eds) Bayesian Statistics 4. Oxford University Press, London, pp 35–60
Berger JO, Pericchi LR (1996) The intrinsic Bayes factor for model selection and prediction. J Amer Stat Assoc 91:109–122
Berger JO, Sellke T (1987) Testing a point null hypothesis: the irreconcilibility of p-values and evidence (with discussion). J Amer Stat Assoc 82:112–122
Berger JO, De Oliveira V, Sansó B (2001) Objective Bayesian analysis of spatially correlated data. J Amer Stat Assoc 96:1361–1374
Carlin BP, Gelfand AE, Smith AFM (1992). Hierarchical Bayesian analysis of change point problems. Appl Stat 41:389-405
Carlstein E (1988) Nonparametric change-point estimation. Ann Stat 16:188–197
Casella G, Berger RL (1987). Reconciling Bayesian and frequentist evidence in the one-sided testing problem (with discussion). J Amer Stat Assoc 82:106–111
Casella G, Moreno E (2002) Objective Bayesian variable selection. Technical Report, University of Granada
Casella G, Moreno E (2003) Objective Bayesian analysis of contingency tables. Technical Report, University of Granada
Casella G, Moreno E (2005) Intrinsic meta analysis of contingency tables. Stat Med 24:583–604
Chernoff H, Zacks S (1964) Estimating the current mean of a normal distribution which is subjected to changes in time. Ann Math Stat 35:999–1018
Choy JH, Broemeling LD (1980) Some Bayesian inferences for a changing linear model. Technometrics 22:71–78
Clyde M. (2001) Discussion to Chipman H, George E, McCulloch RE (2001). IMS Lecture Notes-Monograph Series 38:67–134
Cobb GW (1978) The problem of the Nile: conditional solution to a change-point problem. Biometrika 65:243–251
Dümbgen L (1991) The asymptotic behavior of some nonparametric change-point estimators. Ann Stat 19:1471–1495
Ferreira PE (1975) A Bayesian analysis of a switching regression model: known number of regimes. J Amer Stat Assoc 70:370–374
Hsu DA (1979) Detecting shifts in parameters in gamma sequences with application to stock price and air traffic flow analysis. J Amer Stat Assoc 74:31–40
Jarret RG (1979) A note on the intervals between coal-mining disasters. Biometrika 66:191–193
Jeffreys H (1961) Theory of probability. Oxford University Press, London
Kim S, Sun D (2000). Intrinsic priors for model selection using an encompassing model. Life Time Data Anal 6:251–269
Lai TL (1995) Sequential change point detection in quality control and dynamical systems. J R Stat Soc Ser B 57:613–658
Lorden G (1971) Procedures for reacting to a change in distribution. Ann Math Stat 41:520–527
Menzefrike U (1981) A bayesian analysis of a change in the precision for a sequence of independent normal random variables at an unknown time point. Appl Stat 30:141–146
Moreno E, Liseo B (2003) Default prior for testing the number of components of a mixture. J Stat Plan Inference 111:129–142
Moreno E, Bertolino F, Racugno W (1998) An intrinsic limiting procedure for model selection and hypothesis testing. J Amer Stat Assoc 93:1451–1460
Moreno E, Bertolino F, Racugno W (1999) Default Bayesian analysis of the Behrens-Fisher problem. J Stat Plan Inference 81:323–333
Moreno E, Bertolino F, Racugno W (2000) Bayesian model selection approach to analysis of variance under heterocedasticity. J R Stat Soc Ser D 49:1–15
Moreno E, Bertolino F, Racugno W (2003) Bayesian inference under partial prior information. Scand J Stat 30:565–580
Morris CN (1987) Discussion of Berger/Sellke and Casella/Berger. J Amer Stat Assoc 82:112–122
Müller HG (1992). Change-points in nonparametric regression analysis. Ann Stat 20:737–761
Page ES (1954) Continuous inspection schemes. Biometrika 41:100–114
Page ES (1955) A test for a change in a parameter occurring at an unknown point. Biometrika 42:523–527
Pettitt AN (1979) A non-parametric approach to the change-point problem. Appl Stat 28:126–135
Pollak M, Siegmund D (1991) Sequential detection of a change in a normal mean when the initial value is unknown. Ann Stat 19:394–416
Raftery AE, Akman VE (1986) Bayesian analysis of a Poisson process with a change-point. Biometrika 73:85–89
Rudemo M (1982) Empirical choice of histograms and kernel density estimators. Scand J Stat 9:65–78
Sen AK, Srivastava MS (1973) On multivariate test for detecting change in mean. Sankhy A 35:173–186
Siegmund D (1988) Confidence sets in change point problems. Int Stat Rev 56:31–48
Smith AFM (1975) A Bayesian approach to inference about a change-point in a sequence of random variables. Biometrika 62:407–416
Smith AFM, Cook DG (1980) Straight lines with a change point: a Bayesian analysis of some renal ransplant data. Appl Stat 29:180–189
Sweeting TJ (2001) Coverage probability bias, objective Bayes and the likelihood principle. Biometrika 88:657–675
Wasserman L (2000) Asymptotic inference for mixture models using data-dependent priors. J R Stat Soc Ser B 62:159–180
Worsley KJ (1986) Confidence regions and tests for a change-point in a sequence of exponential random variables. Biometrika 73:91–104
Acknowledgements
We are grateful to two anonymous referees for their comments which have improved an earlier version of the paper. This work has been partially supported by Ministerio de Educación y Ciencia under grant SEJ2004-02447 and BEC2002-00081.
Author information
Authors and Affiliations
Corresponding author
Appendices
Appendix 1
Proof of Lemma 1.
For the models \(M_0 :P\left( {x_1 |\theta } \right) = {{\theta ^{x_1 } } \mathord{\left/ {\vphantom {{\theta ^{x_1 } } {x_1 !}}} \right. \kern-\nulldelimiterspace} {x_1 !}}\,\exp \left\{ { - \theta } \right\},\) and \(M_1 :\left\{ {P\left( {x_1 |\theta _1 } \right)P\left( {x_2 |\theta _2 } \right),\pi ^D \left( {\theta _1 ,\theta _2 } \right) = k\theta _1^{ - 1/2} \theta _2^{ - 1/2} } \right\},\) where θ is an arbitrary but fixed value, and k is an arbitrary positive constant. The minimal training sample is a pair of independent random variables X1,X2 such that under model M1, X i ∽ P(x i |θ i ), and under M0, X i ∽ P(x i |θ), i=0,1. Then, simple calculations give
Furthermore,
Using the equality \(\sum\nolimits_{x = 0}^\infty {{{\left( {\theta \;\theta _1 } \right)^{x_2 } } \mathord{\left/ {\vphantom {{\left( {\theta \;\theta _1 } \right)^{x_2 } } {\Gamma \left( {x + 1/2} \right)x!}}} \right. \kern-\nulldelimiterspace} {\Gamma \left( {x + 1/2} \right)x!}} = {{F_0^1 \left( {1/2,\theta \;\theta _1 } \right)} \mathord{\left/ {\vphantom {{F_0^1 \left( {1/2,\theta \;\theta _1 } \right)} {\Gamma \left( {1/2} \right)}}} \right. \kern-\nulldelimiterspace} {\Gamma \left( {1/2} \right)}}} \) and then substitution in Eq. 12, Lemma 1 follows.
Appendix 2
Proof of Lemma 2.
Consider the model
and
The minimal training sample is a random vector (X1,X2,Y1,Y2) with independent components such that under model M1, X i ~ N(x i |μ 1,σ 21 ), Y i ~ N(y i |μ 2,σ 22 ), and under M *0 , X i , Y i ~ N(x|θ ,τ 2), i=1,2. We recall that a minimal training sample is a random vector of minimal size for which the marginal density is greater than zero and finite (except for a null set with respect to the Lebergue measure). Then,
where
Therefore,
where
Changing to the new variables
the result in Lemma 1 follows.
Rights and permissions
About this article
Cite this article
Moreno, E., Casella, G. & Garcia-Ferrer, A. An objective Bayesian analysis of the change point problem. Stoch Environ Res Ris Assess 19, 191–204 (2005). https://doi.org/10.1007/s00477-004-0224-2
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00477-004-0224-2