Previous |  Up |  Next

Article

Keywords:
multinomial distribution; algorithms; goodness-of-fit divergence tests; power divergence statistics; chi-squared tests; power comparisons
Summary:
In this paper we introduce several algorithms to generate all the vectors in the support of a multinomial distribution. Computational studies are carried out to analyze their efficiency with respect to the CPU time and to calculate their efficiency frontiers. The proposed algorithm is used to calculate exact distributions of power divergence test statistics under the hypothesis of uniformity. Finally, several exact power comparisons are done for different divergence statistics and families of alternatives to the uniformity hypothesis.
References:
[1] Aho A. V., Hopcroft J. E., Ullman J. D.: Data Structures and Algorithms. Addison–Wesley, Massachusetts 1983 MR 0666695 | Zbl 0487.68005
[2] Ali S. M., Silvey S. D.: A general class of coefficient of divergence of one distribution from another. J. Roy. Statist. Soc. Ser. B 286 (1966), 131–142 MR 0196777
[3] Cressie N. A. C., Read T. R. C.: Multinomial goodness of fit tests. J. Roy. Statist. Soc. Ser. B 46 (1984), 440–464 MR 0790631 | Zbl 0571.62017
[4] Csiszár I.: Eine Informationstheoretische Ungleichung und ihre Anwendung auf den Beweis der Ergodizität von Markoffschen Ketten. Publ. Math. Inst. Hungarian Academy of Sciences, Series A, 8 (1963), 85–108 MR 0164374
[5] Kulmann H.: Notes on the computation of the exact distribution function of the $\chi ^2$ and related tests statistics in the equiprobable case. Comput. Stat. Data Anal., The Statistical Software Newsletter 4 (1996), 707–710
[6] Liese F., Vajda I.: Convex Statistical Distances. Teubner, Leipzig 1987 MR 0926905 | Zbl 0656.62004
[7] Marhuenda M. A., Marhuenda, Y., Morales D.: Algorithms to calculate the exact distribution function of power divergence statistics. Technical Report of the Operational Research Center, Miguel Hernández University of Elche 2001
[8] Read T. R. C., Cressie N. A. C.: Goodness–of–fit Statistics for Discrete Multivariate Data. Springer–Verlag, New York 1988 MR 0955054 | Zbl 0663.62065
[9] Weiss M. A.: Data Structures and Algorithm Analysis. Benjamin/Cummings Publishing Company, Redwood City, CA 1992 Zbl 0879.68016
[10] Zografos K., Ferentinos, K., Papaioannou T.: $\phi $-divergence statistics: sampling properties, multinomial goodness of fit and divergence tests. Comm. Statist. A – Theory Methods 19 (1990), 1785–1802 DOI 10.1080/03610929008830290 | MR 1075502
Partner of
EuDML logo