Estimating a discrete distribution via histogram selection
ESAIM: Probability and Statistics, Tome 15 (2011), pp. 1-29.

Our aim is to estimate the joint distribution of a finite sequence of independent categorical variables. We consider the collection of partitions into dyadic intervals and the associated histograms, and we select from the data the best histogram by minimizing a penalized least-squares criterion. The choice of the collection of partitions is inspired from approximation results due to DeVore and Yu. Our estimator satisfies a nonasymptotic oracle-type inequality and adaptivity properties in the minimax sense. Moreover, its computational complexity is only linear in the length of the sequence. We also use that estimator during the preliminary stage of a hybrid procedure for detecting multiple change-points in the joint distribution of the sequence. That second procedure still satisfies adaptivity properties and can be implemented efficiently. We provide a simulation study and apply the hybrid procedure to the segmentation of a DNA sequence.

DOI : 10.1051/ps/2009007
Classification : 62G05, 62C20, 41A17
Mots clés : adaptive estimator, approximation result, categorical variable, change-point detection, minimax estimation, model selection, nonparametric estimation, penalized least-squares estimation
@article{PS_2011__15__1_0,
     author = {Akakpo, Nathalie},
     title = {Estimating a discrete distribution \protect\emph{via} histogram selection},
     journal = {ESAIM: Probability and Statistics},
     pages = {1--29},
     publisher = {EDP-Sciences},
     volume = {15},
     year = {2011},
     doi = {10.1051/ps/2009007},
     language = {en},
     url = {http://archive.numdam.org/articles/10.1051/ps/2009007/}
}
TY  - JOUR
AU  - Akakpo, Nathalie
TI  - Estimating a discrete distribution via histogram selection
JO  - ESAIM: Probability and Statistics
PY  - 2011
SP  - 1
EP  - 29
VL  - 15
PB  - EDP-Sciences
UR  - http://archive.numdam.org/articles/10.1051/ps/2009007/
DO  - 10.1051/ps/2009007
LA  - en
ID  - PS_2011__15__1_0
ER  - 
%0 Journal Article
%A Akakpo, Nathalie
%T Estimating a discrete distribution via histogram selection
%J ESAIM: Probability and Statistics
%D 2011
%P 1-29
%V 15
%I EDP-Sciences
%U http://archive.numdam.org/articles/10.1051/ps/2009007/
%R 10.1051/ps/2009007
%G en
%F PS_2011__15__1_0
Akakpo, Nathalie. Estimating a discrete distribution via histogram selection. ESAIM: Probability and Statistics, Tome 15 (2011), pp. 1-29. doi : 10.1051/ps/2009007. http://archive.numdam.org/articles/10.1051/ps/2009007/

[1] M. Aerts and N. Veraverbeke, Bootstrapping a nonparametric polytomous regression model. Math. Meth. Statist. 4 (1995) 189-200. | MR | Zbl

[2] Y. Baraud and L. Birgé, Estimating the intensity of a random measure by histogram type estimators. Prob. Theory Relat. Fields 143 (2009) 239-284. | MR | Zbl

[3] A. Barron, L. Birgé and P. Massart, Risk bounds for model selection via penalization. Prob. Theory Relat. Fields 113 (1999) 301-413. | MR | Zbl

[4] C. Bennett and R. Sharpley, Interpolation of operators, volume 129 of Pure and Applied Mathematics. Academic Press Inc., Boston, M.A. (1988). | MR | Zbl

[5] L. Birgé, Model selection via testing: an alternative to (penalized) maximum likelihood estimators. Ann. Inst. H. Poincaré Probab. Statist. 42 (2006) 273-325. | Numdam | MR

[6] L. Birgé, Model selection for Poisson processes, in Asymptotics: Particles, Processes and Inverse Problems, Festschrift for Piet Groeneboom. IMS Lect. Notes Monograph Ser. 55. IMS, Beachwood, USA (2007) 32-64. | MR | Zbl

[7] L. Birgé and P. Massart, Minimal penalties for Gaussian model selection. Prob. Theory Relat. Fields 138 (2007) 33-73. | MR | Zbl

[8] J.V. Braun and H.-G. Müller, Statistical methods for DNA sequence segmentation. Stat. Sci. 13 (1998) 142-162. | Zbl

[9] J.V. Braun, R.K. Braun and H.-G. Müller, Multiple changepoint fitting via quasilikelihood, with application to DNA sequence segmentation. Biometrika 87 (2000) 301-314. | MR | Zbl

[10] T.H. Cormen, C.E. Leiserson, R.L. Rivest and C. Stein, Introduction to algorithms. Second edition. MIT Press, Cambridge, MA (2001). | MR | Zbl

[11] M. Csűrös, Algorithms for finding maximum-scoring segment sets, in Proc. of the 4th international workshop on algorithms in bioinformatics 2004. Lect. Notes Comput. Sci. 3240. Springer, Berlin, Heidelberg (2004) 62-73. | MR

[12] R.A. Devore and G.G. Lorentz, Constructive approximation. Springer-Verlag, Berlin, Heidelberg (1993). | MR | Zbl

[13] R.A. Devore and R.C. Sharpley, Maximal functions measuring smoothness. Mem. Amer. Math. Soc. 47 (1984) 293. | MR | Zbl

[14] R.A. Devore and X.M. Yu, Degree of adaptive approximation. Math. Comp. 55 (1990) 625-635. | MR | Zbl

[15] C. Durot, E. Lebarbier and A.-S. Tocquet, Estimating the joint distribution of independent categorical variables via model selection. Bernoulli 15 (2009) 475-507. | MR | Zbl

[16] Y.-X. Fu and R.N. Curnow, Maximum likelihood estimation of multiple change points. Biometrika 77 (1990) 562-565. | MR | Zbl

[17] S. Gey S. and E. Lebarbier, Using CART to detect multiple change-points in the mean for large samples. SSB preprint, Research report No. 12 (2008).

[18] M. Hoebeke, P. Nicolas and P. Bessières, MuGeN: simultaneous exploration of multiple genomes and computer analysis results. Bioinformatics 19 (2003) 859-864.

[19] E. Lebarbier, Quelques approches pour la détection de ruptures à horizon fini. Ph.D. thesis, Université Paris Sud, Orsay, 2002.

[20] E. Lebarbier and E. Nédélec, Change-points detection for discrete sequences via model selection. SSB preprint, Research Report No. 9 (2007).

[21] P. Massart, Concentration inequalities and model selection. Lectures from the 33rd Summer School on Probability Theory held in Saint-Flour, July 6-23, 2003. Lect. Notes Math. 1896. Springer, Berlin, Heidelberg (2007). | MR | Zbl

[22] P. Nicolas et al., Mining Bacillus subtilis chromosome heterogeneities using hidden Markov models. Nucleic Acids Res. 30 (2002) 1418-1426.

[23] W. Szpankowski, L. Szpankowski and W. Ren, An optimal DNA segmentation based on the MDL principle. Int. J. Bioinformatics Res. Appl. 1 (2005) 3-17.

Cité par Sources :