With the increasing use of fossil fuels since the industrial revolution, greenhouse gas emission values have also been increasing over the years. Increasing greenhouse gases is one of the major causes of global warming, which is one of today's most important problems. In the study, clustering analysis was performed using 1990-2014 values according to 15 indicators in the World Bank data of 36 countries that are members of the organisation for Economic Co-operation and Development (OECD). The data was evaluated using data mining methods and two-stage clustering analysis due to the large number of total data. At the time of application, the ideal number of clusters was found to be four. Clustering results were obtained separately using SPSS, MATLAB and WEKA programs. Two-stage and self-organizing map (SOM) type neural networks are preferred as clustering methods. For the obtained cluster results, the countries have been combined for taking into account the recurring cluster values. As a result, countries are clustered so that the carbon emission values used are ranked from high to low. The number of countries in the cluster was 1 in the first cluster, 9 countries in the second cluster, 10 countries in the the third cluster and 16 countries in the fourth cluster. Clustering results may guide programs to reduce carbon emissions to countries in the same cluster. In addition, commissions including the countries in the related cluster can be established under the OECD union. The fact that climate change has made him feel more and more recently supports the implementation of measures and practices as soon as possible.
Keywords: SOM Neural Networks, Data Mining, Two Stage Clustering, Carbon Emission, Ideal Cluster Number