Analyzing Spatial Structure of Frontier Cities in Western Azarbaijan through Multi-variety Statistical Analysis and Neural Networks

Extended Abstract
Introduction
This study investigates spatial structure of frontier cities in Western Azarbaijan and effective factors on it by taking into consideration population, social, economic, cultural, health, infrastructure, transportation, and communication and body indices. A brief look at literature on applied and theoretical issues in developing borderline cities indicates three approaches location traditional approach, cross- border cooperation approach, people approach. christaller, losch, Giersch, Perrouy, Friedman are included among those who has dealt with the first approach. The second approach was integrated into economic geography from 1990. This approach does borderline analysis with regard to global city theories, globalization, confidence, cooperation and unity. Based on this approach views borderline areas the same from different aspect like geographical features, history, ethnic groups, and economic chances. This approach confirm that its only political and governmental affairs that cut off the two areas. The third approach views borderline areas as integration. Scholar analyze borderline areas and its stability with regard to birth and increase in population.

Methodology
Methodology is descriptive-analytical and co- relational. Qualitative planning models such as Top sis multi-criteria decision-making, Entropy, dispersion coefficient, are used for ranking, and neural network, regression analysis, route analysis and variance analysis are used to analyze data.

Results and Discussion
Based on administrative – political division province of west Azarbaijan included 36 cities among them we find 13 borderline cities. Borderline Cities are those cities located less than 50 km the border. The result shows that among the 13 cities of Western Azarbaijan, based on political - administrative division in 2007, Piranshahr, Sardasht and Oshnavieh cities which comprise 67/8 percent of population of frontier cities are classified as developed frontier cities. Cities of Siah cheshmeh, Avajigh, Gardan keshaneh, Tazeh shahr and Poldasht which include 18/7 percent of population in frontier cities are classified as semi-developed and Mir abad, Silvaneh, Bazargan, Rabt and Sarv cities which have 13/5 percent of population in frontier cities are considered as under-developed cities. Dispersion coefficient shows that the highest difference among indices is related to cultural index and the lowest one is social index.

Conclusion
According to the results of route analysis, cultural section has had the highest and infrastructure section has had the lowest effect on spatial structure of frontier cities. Neural network analysis shows that infrastructure section with a rate of 100 percent and population section with a rate of 8/8 percent have had the most and the least significance in spatial structure of frontier cities. Degree of inequality (difference) between rank of cities with respect to different sections was examined through multi-variants analysis and the result showed that the development of cities in different indices have not been equal. Therefore application of urban development strategies resulting from statistical analysis models based on rank order of cities seems necessary.