Typological Grouping Based on Decomposition of Probability Distributions Mixtures
https://doi.org/10.34020/2073-6495-2020-1-255-267
Abstract
A mixture of probability distributions is a mathematical model that allows to describe heterogeneous data. The task of separating mixtures or decomposition is the task of estimating the unknown parameters of miscible distributions. Despite the adequacy of the description of heterogeneous data, the decomposition of mixtures is a separate problem, due to the large number of parameters to be evaluated. The article carries out historical periodization,systematization, and a critical comparative analysis of existing methods and algorithms for decomposition of mixtures of probability distributions, identifies the possibilities and limitations of their application for the analysis of real populations. Based on existing algorithms, a method for separating mixtures of an arbitrary known number of probability distributions and a further typological grouping of real socio-economic aggregates is proposed. Unlike existing methods, a method for calculating threshold values to determine the boundaries of types and the number of components of the mixture, in cases where it is unknown, is proposed. Based on the proposed methodology, a typology of the subjects of the Russian Federation by the level of unemployment in the Russian Federation is carried out.
About the Authors
Yu. N. Ismaiylova
Novosibirsk State University of Economics and Management
Russian Federation
Ismaiylova Yuliya N., Senior Lecturer, Department of Statistics
Russian Federation
Ismaiylova Yuliya N., Senior Lecturer, Department of Statistics
S. E. Khrushchev
Novosibirsk State University of Economics and Management
Russian Federation
Khrushchev Sergey E., Candidate of Physical and Mathematical Sciences, Associate Professor, Department of Statistics
Russian Federation
Khrushchev Sergey E., Candidate of Physical and Mathematical Sciences, Associate Professor, Department of Statistics
References
1. Aitkin M., Aitkin I. Efficient computation of maximum likelihood estimates in mixture distributions, with reference to overdispersion and variance components. In Proceedings XVIIth International Biometric Conference, Hamilton, Ontario. Alexandria, Virginia: Biometric Society. 1994. P. 123–138.
2. Akaike Н. Information theory and an extension of the maximum likelihood principle. in: B.N. Petrov and F. Csake (eds.) Second International Symposium on Information Theory. Akademiai Kiado, Budapest, 1973. P. 267–281.
3. Akaike Н. A Bayesian analysis of the minimum AIC procedure. Ann. Inst. Statist. Math. 1978.Vol. 30A. P. 9–14.
4. Blischke W.R. Estimating the parameters of mixtures of binomial distributions. J. Amer. Statist. Assoc. 1964 – 59. № 306. P. 510–528.
5. Blischke W.R. Moment estimators for the parameters of a mixture of two binomial distributions. Ann. Math. Stat. 1962 – 33. № 2. P. 444–454.
6. Cohen A.C. Estimation in mixtures of discrete distributions. Proc Int Symp. Classical and Contagious Discrete Distrib. Montreal, 1963. P. 373–378.
7. Dempster A.P., Laird N.M., Rubin D.B. Maximum likelihood from incomplete data via the EM algorithm (with discussion). Journal of the Royal Statistical Society B. 1977 – 39. P. 1–38.
8. Glinskiy V., Serga L., Chemezova E., Zaykov K. Clusterization economy as a way to build sustainable development of the region. Procedia CIRP 13. «13th Global Conference on Sustainable Manufacturing – Decoupling Growth from Resource Use». 2016. P. 324–328.
9. Glinskiy V., Serga L., Khvan M. Assessment of environmental parameters impact on the level of sustainable development of territories. Procedia CIRP 13. «13th Global Conference on Sustainable Manufacturing – Decoupling Growth from Resource Use». 2016. P. 626–631.
10. Joffe A.D. Mixed exponential estimation by the method of half moments. Appl. Statist. 1964 – 13. № 2. P. 91–98.
11. Lange K. A quasi-Newton acceleration of the EM algorithm. Statistica Sinica. 1995. 5. P. 1–18.
12. McLachlan, Geoffrey J., Krishnan Thriyambakam, Ng, See Ket. The EM Algorithm, Papers / Humboldt-Universität Berlin, Center for Applied Statistics and Economics (CASE), 2004, 24.
13. Redner R.A., Walker H.E. Mixture densities, maximum likelihood and the EM algorithm. SIAM Review. 1984. 26. P. 195–239.
14. Tallis G.M., Light R. The use of fractional moments for estimating the parameters of a mixed exponential distribution.Teehnometniics. 1968 – 10. № 1. P. 161–175.
Review
For citations:
Ismaiylova Yu.N., Khrushchev S.E. Typological Grouping Based on Decomposition of Probability Distributions Mixtures. Vestnik NSUEM. 2020;(1):255-267. (In Russ.) https://doi.org/10.34020/2073-6495-2020-1-255-267
Email this article 
