TR
Avrasya Ekonometri �statistik ve Ampirik Ekonomi DergisiYl:2015 Say: 2 Alan: statistik

Volkan Sevin
TRK?YE?DEK? BANKALARIN KARI?IM MODELLER?NE DAYALI BAYESC? KMELEME ANAL?Z? ?LE SINIFLANDIRILMASI
 
Bankaclk sektr bir lkenin ekonomisinin nemli paralarndan biridir. Bankalar arasnda baz ekonomik ve finansal gstergeler bakmndan farklar ve benzerlikler vardr. Bankaclk sektrne ve yatrmclara bir bavuru kayna olarak bankalarn benzerlikleri ve farklar bakmndan snflandrlmalar yararl olacaktr. Bu almada, Trkiyedeki bankalarn baz ekonomik ve finansal gstergelere gre snflandrlmas amalanmtr. Birok snflandrma almas geleneksel deterministik yntemlere dayanmaktadr. Ancak, model bazl kmeleme analizi, snflandrma ilemine olaslksal bir yaklam da katmaktadr. Karm modelleri, model bazl kmeleme analizinde geni bir ekilde kullanlmaktadr. Karm modellerinin Bayes tahmini, literatrde grece olarak yenidir. Bu nedenle bu aratrmada, karm modellerine dayal bir Bayesci kmeleme analizi gerekletirilmi ve Trkiyedeki bankalarn bir snflandrmas yaplmtr. Sonu ksmnda bankalarn oluturduu gruplar tablolar halinde sergilenmi ve gruplar hakknda yorumlar yaplmtr.

Anahtar Kelimeler: Kmeleme analizi, Bayesci kmeleme analizi, Karm modelleri, Trk bankaclk sektr


A CLASSIFICATION OF THE BANKS IN TURKEY WITH BAYESIAN CLUSTER ANALYSIS BASED ON MIXTURE MODELS
 
Banking sector is an important part of the economy of a country. There are differences and similarities among these banks in terms of some economic and financial indicators. It would be useful to make a classification research about these banks regarding their differences and similarities to provide a reference to the banking sector and investors. In this research, it is aimed to classify the banks in Turkey with respect to several financial and economic indicators. There are two main kind of clustering methods in the literature: Deterministic versus Model-based cluster analysis. Most clustering studies are based on traditional deterministic methods. However, model-based cluster analysis provides a probabilistic approach to clustering. Mixture models are widely used in model based cluster analysis. Bayesian estimation of mixture models in the literature is relatively new. Therefore, in this research, Bayesian cluster analysis based on mixture models is employed and a classification of the banks is made using the selected method.

Keywords: Cluster analysis, Bayesian cluster analysis, Mixture models, Turkish banking sector.


Detay

ÇERK