Use of linear programming model to construct the nomenclature of territorial units for statistics (NUTS): An application for Turkey

Geographic classifications are used to create statistics that shed light on regional development policies. Legislation has defined the fundamental principles and criteria governing the European Union geographical classification system, NUTS. When comparing NUTS in the EU and NUTS-2002 in our country, both classifications are incompatible with legislation when using up-to-date population data. In our country, however, incompatibility is far more serious. This inconsistency has a significant impact on the quality and efficiency of the regional statistics produced. It also has a significant impact on the success of regional policies. In this study, a set of geographical classification data will be produced to generate statistical information that will shed light on long-term and sustainable regional development policies, in accordance with relevant legislation, by adhering to predetermined basic criteria about country borders and administrative structures of countries. The study designs geographical classification as an optimization problem, a p-median problem, which is a subset of location allocation problems. By creating an integer programming model for the problem at hand, the problem solution set is sought. The grid code of 786470 grids, population, latitude, longitude, x and y coordinates, altitude, district and city names, and codes to which the grid center is connected were used as the main data source in the study's application. The newly created NUTS and NUTS-2002 were compared as a result of the application. As a result, the previously observed 100% non-compliance with NUTS legislation in NUTS-2002 has been transformed into 100% compliance in the newly created NUTS for Turkey.