Hybrid methodology of panel data analysis used on gross domestic product of various countries


Dönmez S., Özaydın Ö.

8 th International Conference on Advances in Statistics, Gazimagusa, Cyprus (Kktc), 16 - 18 May 2022

  • Publication Type: Conference Paper / Unpublished
  • City: Gazimagusa
  • Country: Cyprus (Kktc)
  • Eskisehir Osmangazi University Affiliated: Yes

Abstract

We used a hybrid methodology of linear regression on panel data and panel data clustering on gross domestic product per year of various countries between 1968-2015. Our aim was to see the difference between the growth trends of the countries and evaluate their numerically expressed growths according to their significance. The countries used in the data are Afghanistan, Albania, Argentina, Australia, Austria, Burundi, Belgium, Benin, Burkina Faso, Bangladesh, Bulgaria, Bolivia, Brasil, Barbados, Botswana, Central African Republic, Canada, Switzerland, Chile, China, Cote d’Ivor, Cameroon, Congo, Democratic Republic of Congo, Comoros, Cabo Verde, Costa Rica, Czech Republic, Cyprus, Germany, Djibouti, Dominica, Dominican Republic, Equador, Egypt, Ethiopia, Spain, France, Finland, Gabon, United Kingdom, Ghana, Guinea, Guinea-Bissau, Greece, Guatemala, Hong Kong and Honduras. Even though these countries can generally be categorized into developing and developed countries, we wanted to evaluate the clustering algorithm on the basis of coming up with the amount of clusters just enough to make a meaningful distinction between countries other than the aforementioned subjective categorization. Our findings showed us that all countries can be expressed in 6 clusters even though there are minor differences between the clusters.