كاربرد تركيب الگوريتم خوشه بندي و الگوريتم رقابت استعماري (ICA) در سطح-بندي توسعه يافتگي مناطق روستايي (مطالعه موردي: بخش مركزي شهرستان بويراحمد)

Application of Clustering Algorithm and Imperialist Competitive Algorithm (ICA) in Zoning of Rural Areas: (Case study: Central District of Boyer-Ahmad County(

دراثر برنامه ريزي هاي نامطلوب و متمركز گذشته، كيفيت توسعه و زيرساخت هاي آن مسايل عمده اي را در روند توسعه روستاهاي كشور ايجاد كرده است. ابعاد گوناگون و پيچيدگي ساختاري اين موضوع، يكي از تنگناهاي اساسي در ارايه مدل مناسب براي توزيع اعتبارات به شمار مي آيد. دراين راستا، در پژوهش حاضر به موضوع سطح بندي مناطق روستايي بخش مركزي شهرستان بويراحمد مي پردازيم. اين تحقيق، به لحاظ هدف كاربردي است. جامعه آماري پژوهش، روستاهاي بالاي 20 خانوار و بخش مركزي شهرستان بويراحمد (89 روستا) بودند كه 566 روستايي مطالعه شدند. براي جمع آوري داده ها، علاوه بر اسناد، از پرسش نامه محقق ساخته نيز استفاده شد. ضريب آلفاي كرونباخ (561/0- 955/0) به دست آمده، بيانگر بهينگي پرسش نامه بود. براي تعيين شاخص هاي به كاربرده شده، از رويكرد اندام وار استفاده شد. براي تعيين سطوح توسعه يافتگي، از روش تحليل خوشه اي به شيوه K-means و الگوريتم رقابت استعماري (ICA) استفاده شد. داده پردازي نيز با استفاده از نرم افزار MATLAB 7.10 انجام شد. يافته ها نشان داد كه جواب بهينه الگوريتم تركيبي K-means و ICAاز بهترين جواب الگوريتم K-means بهتر است؛ اين موضوع اعتبار اين الگوريتم را به وضوح نشان مي دهد. نتايج پژوهش حاكي از آن است كه پراكنش فضايي توسعه يافتگي در منطقه موردمطالعه در وضعيت متوازني قرار ندارد؛ بهطوري كه اغلب روستاها (43/49%) در سطح نسبتاً محروم قرار مي گيرند. همچنين، در سطوح در حال توسعه و توسعه يافته به ترتيب، 08/28% و 47/22% از روستاها قرار دارند. افزون براين، روستاهاي دهستان سررود شمالي از وضعيت توسعه-يافتگي بهتري نسبت به ساير روستاهاي بخش مورد مطالعه برخوردار هستند.

1. INTRODUCTION Due to unfavorable and concentrated planning, the quality of development and its infrastructures have brought about major problems for the development process of the villages in Iran. Various dimensions of development and its structural complexity impose major limitations on presentation of an appropriate model for the budget allocation. Therefore, awareness of distribution of development in rural areas is essential for planning and rural development programs. Grading the levels of development is a powerful instrument for identifying and increasing our knowledge about such areas and management of their development. This study aims at determining levels of development in the Central district of Boyer-Ahmad county. Therefore, given the primary aim of the study, namely using a suitable model to determine levels of development and homogenization in the study area, we have used a compound model of clustering algorithm with k-means method and Imperialist Competitive Algorithm. 2. METHODOLOGY Given the aim of the study, this is an applied research which has employed quantitative data analysis method. Statistical population of the study included the villages in the central district of Boyer-Ahmad County having 20 or more households (89 villages). The sample size was determined by Lin sampling table. Randomly selected from the population, final sample included 566 rural households. A questionnaire was devised by the researcher to collect the data. To test validity of research tool, Cronbachʹs alpha was used. The coefficient of Cronbachʹs alpha (0.610-0.864) in turn suggested that the questionnaire was optimal. To formulate a logical order of indices which represents features of rural development, based on literature, the key indices were identified and selected in a systematic way. 51 criteria were identified and finalized. In order to determine the zoning of rural development in the Central district of Boyer-Ahmad county, descriptive analysis and K-means algorithm together with imperialist competitive algorithm (ICA) were used. Moreover, Arc-GIS9.3 software was instrumental in mapping the spatial distribution of development. Data processing was performed using MATLAB 7.10. 3. DISCUSSION Compared to mathematical methods, intelligent optimization algorithm approach shows significant progress. Parameters that should be set during the mentioned algorithm, are: ، ، ، and . Remarkably, in this method, data sets are insensitive to changes in parameters. Based on these findings, the optimal solution algorithm, K-MICA, is 7848, but the K-means algorithm, under no circumstances is able to reach this amount. Furthermore, the worst response from K-MICA algorithm is better than the best response from K-means algorithm, and it clearly shows the validity of this algorithm. The results indicate that development of the study area is so unfavorable that most of the villages (43.49 %) having 19.63 % of the total population are in relatively disadvantaged level. 28/08 % of the villages having 22.67 % of the total population were in developing level, and 22.47 % of the villages having 57.70 % of the total population were in developed level. 4. CONCLUSION This study presents a new method for zoning of development in rural and urban areas. The study results suggest that the optimal solution of K-MICA algorithm is better than that of k-means algorithm. Nevertheless, the worst result of K-MICA algorithm is better than the best result of k-means algorithm. Therefore, for zoning of rural and urban areas, it is better to use compound method of K-MICA algorithm rather than traditional clustering. In addition, the study results shows that the study area is mainly undeveloped, and most villages (49.43%) comprising 19.63% of the general population are in poor conditions and in relatively underdeveloped level.