Geodesic
Based on the relative similarity coefficients ranking characteristics forming the cluster partition
Prikladnaya Diskretnaya Matematika. Supplement, no. 10 (2017), pp. 160-162.

Voir la notice de l'article provenant de la source Math-Net.Ru

Assume that cluster partition of the finite set of objects is carried out by the establishing the proximity degree for some numerical characteristics that we call forming ones. We study the problem of estimating a relative informative value of the forming characteristics for the clusterization process. To range them, we introduce a coefficient of the relative cluster strength by which we can estimate a relative value of each of the characteristics in the cluster constructing with the respect to the arbitrary collective of the others. Also another two coefficients are proposed for estimating the degree of so-called cluster connection between any two of the forming characteristics. This connection is understood as the possibility of one characteristic to replace the other without any major changes in the cluster partition. The proposed coefficients represent two different approaches to estimating the strength of the cluster connection. An algorithm for the dimension reduction which allows minimal distortion of the cluster structure based on these coefficients and their modifications is discussed. A distortion degree is considered with the respect to some cluster metric proposed earlier by one of the authors. More confident detection of computer security threats while lowering the total load on the system can be achieved through the implementation of this algorithm.
Keywords: cluster partition, dimension reduction, cluster connection, strength degree coefficients. 
@article{PDMA_2017_10_a61,
     author = {S. V. Dronov and E. A. Evdokimov},
     title = {Based on the relative similarity coefficients ranking characteristics forming the cluster partition},
     journal = {Prikladnaya Diskretnaya Matematika. Supplement},
     pages = {160--162},
     publisher = {mathdoc},
     number = {10},
     year = {2017},
     language = {ru},
     url = {http://geodesic.mathdoc.fr/item/PDMA_2017_10_a61/}
}
TY  - JOUR
AU  - S. V. Dronov
AU  - E. A. Evdokimov
TI  - Based on the relative similarity coefficients ranking characteristics forming the cluster partition
JO  - Prikladnaya Diskretnaya Matematika. Supplement
PY  - 2017
SP  - 160
EP  - 162
IS  - 10
PB  - mathdoc
UR  - http://geodesic.mathdoc.fr/item/PDMA_2017_10_a61/
LA  - ru
ID  - PDMA_2017_10_a61
ER  - 
%0 Journal Article
%A S. V. Dronov
%A E. A. Evdokimov
%T Based on the relative similarity coefficients ranking characteristics forming the cluster partition
%J Prikladnaya Diskretnaya Matematika. Supplement
%D 2017
%P 160-162
%N 10
%I mathdoc
%U http://geodesic.mathdoc.fr/item/PDMA_2017_10_a61/
%G ru
%F PDMA_2017_10_a61
S. V. Dronov; E. A. Evdokimov. Based on the relative similarity coefficients ranking characteristics forming the cluster partition. Prikladnaya Diskretnaya Matematika. Supplement, no. 10 (2017), pp. 160-162. http://geodesic.mathdoc.fr/item/PDMA_2017_10_a61/

[1] Dronov S. V., “Odna klasternaya metrika i ustoichivost klasternykh algoritmov”, Izvestiya AltGU, 2011, no. 1/2, 32–35

[2] Dronov S. V., Dementjeva E. A., “A new approach to post-hoc problem in cluster analysis”, Model Assisted Statistics and Applications, 7:1 (2012), 49–65

[3] Zagoruiko N. G., Kutnenko O. A., “Tsenzurirovanie obuchayuschei vyborki”, Vestnik TGU. Upravlenie, vychislitelnaya tekhnika i informatika, 2013, no. 1(22), 66–73