Classifying Two Populations by Bayesian Method and Applications

Abstract

This article proposes some related issues to classification problem by Bayesian method for two populations. They are relationships between Bayes error (BE) and other measures and the results for determining the BE. In addition, we propose three methods to find the prior probabilities that can make to reduce BE. The calculation of these methods can be performed conveniently and efficiently by the MATLAB procedures. The new approaches are tested by the numerical examples including synthetic and benchmark data and applied in medicine and economics. These examples also show the advantages of the proposed methods in comparison with existing methods.

Keywords

Bayes error / Classification / Measure / Posterior / Prior

References

[1] Berger JO. Statistical Decision Theory and Bayesian Analysis. 1985 Berlin: Springer

[2] Devijver PA, Kittler J. Pattern Recognition: A Statistical Approach. 1982 New York: Prentice Hall

[3] Dunn JC. A fuzzy relative of the isodata process and its use in detecting compact well-separated clusters. Cybern. 2008, 3 3 32-57

[4] Ghosh AK, Chaudhuri P, Sengupta D. Classification using kernel density estimates. Technometrics. 2012, 48 1 377-392

[5] Inman HF, Bradley EL. The overlapping coefficient as a measure of agreement between probability distributions and point estimation of the overlap of two normal densities. Commun. Stat. Theory Methods. 1989, 18 10 3851-3874

[6] James I. Estimation of the mixing proportion in a mixture of two normal distributions from simple, rapid measurements. Biometrics. 1978, 5 265-275

[7] Jasra A, Holmes C, Stephens D. Markov chain Monte Carlo methods and the label switching problem in Bayesian mixture modeling. Stat. Sci.. 2005, 12 50-67

[8] Kraft, C.H.: Some conditions for consistency and uniform consistency of statistical procedures. University of California (1955)

[9] Martinez WL, Martinez AR. Computational Statistics Handbook with MATLAB. 2007 Boca Raton: CRC Press

[10] Matusita K. On the notion of affinity of several distributions and some of its applications. Ann. Inst. Stat. Math.. 1967, 19 1 181-192

[11] McLachlan GJ, Basford KE. Mixture Models. Inference and Applications to Clustering. 1988 New York: Dekker

[12] Miller G, Inkret W, Little T, Martz H, Schillaci M. Bayesian prior probability distributions for internal dosimetry. Radiat. Protect. Dosim.. 2001, 94 4 347-352

[13] Nguyentrang T, Vovan T. A new approach for determining the prior probabilities in the classification problem by Bayesian method. Adv. Data Anal. Classif.. 2017, 11 3 629-643

[14] Nielsen F. Generalized Bhattacharyya and Chernoff upper bounds on Bayes error using quasi-arithmetic means. Pattern Recognit. Lett.. 2014, 42 25-34

[15] Pal NR, Bezdek JC. On cluster validity for the fuzzy c-means model. IEEE Trans. Fuzzy Syst.. 1995, 3 3 370-379

[16] Pham-Gia T, Turkkan N, Bekker A. Bounds for the bayes error in classification: a bayesian approach using discriminant analysis. Stat. Methods Appl.. 2007, 16 1 7-26

[17] Pham-Gia T, Turkkan N, Vovan T. Statistical discrimination analysis using the maximum function. Commun. Stat. Simul. Comput.. 2008, 37 2 320-336

[18] Scott DR. Multivariate Density Estimation: Theory Practice and Visualization. 1992 New York: Wiley

[19] Silverman BW. Density Estimation for Statistics and Data Analysis. 1986 Boca Raton: CRC Press

[20] Toussaint G. Some inequalities between distance measures for feature evaluation. Comput. 1972, 21 389-394

[21] Vovan T. $L^1$-distance and classification problem by Bayesian method. J. Appl. Stat.. 2017, 44 3 385-401

[22] Vovan T, Nguyentrang T. Fuzzy clustering of probability density functions. J. Appl. Stat.. 2017, 44 4 583-601

[23] VoVan T, Pham-Gia T. Clustering probability distributions. J. Appl. Stat.. 2010, 37 11 1891-1910

[24] Webb AR. Statistical Pattern Recognition. 2002 2 Hoboken: Wiley