خوشه‌بندي مددجويان بهزيستي به‌منظور شناسايي هدفمند خانواده‌هاي تحت پوشش با استفاده از تكنيك داده‌كاوي

Clustering of the State Welfare Organization of Iran’s Clients to Identify the Supported Families by Using Data Mining Techniques

هدف: امروزه افزايش معضلات اجتماعي و ازدياد مددجويان بهزيستي باعث شده كه تكنيك‌هاي سنتي توانايي ترسيم وضعيت مشخصي از گروه‌هاي دريافت‌كنندهٔ خدمات را نداشته باشند. نبود دسته‌بندي مناسب از افراد باعث شده بهزيستي نتواند خدمات خود را با نيازهاي افراد هم‌سو كند. تكنيك‌هاي خوشه‌بندي مي‌توانند براي دست‌يابي به اين رده‌بندي كمك شاياني كنند. روش‌بررسي: تحقيق حاضر از لحاظ نوع هدف كاربردي و از نظر روش اجرا توصيفي-پيمايشي است. داده‌هاي استفاده شده مربوط به 4155 نفر از مددجويان بخش اجتماعي بهزيستي استان كردستان از سال 86 به بعد است. به‌علاوه براي پياده‌سازي داده‌كاوي از مدل استاندارد CRISP-DM استفاده شد. يافته‌ها: متناسب با يافته‌ها، چهار خوشه به ‌دست آمد و يكي از خوشه‌ها (خوشهٔ 2) به‌عنوان خوشهٔ بهينه انتخاب شد. از ميان ويژگي‌هاي بررسي‌شدهٔ شهرستان سنندج،گروه هدف معلولين،زنان، بي‌سوادان،گروه متأهلين، ساكنين شهر و بعد مستمري اول (يكم)، ويژگي‌هايي بودند كه فراواني بيشتري را در خوشهٔ بهينه داشتند. به‌علاوه وجود ارتباط بين دوبه‌دوي متغيرها بررسي و مشخص شد كه به‌جز متغيرهاي جنسيت و تحصيلات، ساكن شهر يا روستا و بعد مستمري، تحصيلات و ساكن شهر يا روستا ارتباط دوبه‌دويي بين ساير متغيرها وجود دارد. نتيجه‌گيري: باتوجه به يافته‌ها بهزيستي مي‌تواند با تمركز بيشتر در برطرف‌كردن نيازهاي افراد با ويژگي‌هاي ساكن مناطق شهري سنندج، افزايش مستمري معلوليت، جنسيت زن، بي‌سوادي، متأهل‌بودن، خدمات خود را با نيازمندي‌هاي مددجويان هم‌راستا كند.

Background: In recent years public service has become one of the fastest growing sectors of the world economy and is widely recognized for its contribution to regional and national economic development. The exacerbation and growth of social problems and the increasing number of welfare clients has made the traditional techniques inefficient to find the exact and specific information about the needy. Insufficient data about the families and their needs besides the inappropriate categorization for future plans requires data analysis and implementation. To fulfill this important need, Clustering technique in data mining can be useful and helpful. So, this study aims to cluster the clients of State Welfare Organization of Iran so as to identify the supported families for responding the clients' needs in a better way. Methods: This paper follows a practical objective with a descriptive-survey method of research. The Standard Model of CRISP-DM is used to implement data mining. Data mining is the process of discovering the significance of user knowledge such as patterns from large amount of data stored in databases. Very appeal studies have employed data mining to identify the supported families in State Welfare Organization. Also, it is completely unique in Iran. In order to group, predict, recognize and satisfy the needs of the supported individuals, social data of clients of the State Welfare Organization of Iran in Kurdistan province were collected since 1384. Next, a database containing 4155 user’s data with seven attributes were used. The attributes include cities, number of persons supported by The State Welfare Organization, purpose groups, gender, place of living (city/village) attribute, educational degree and finally marriage status. Results: By using Rapid Minder software and applying random clustering technique, four clusters were achieved and cluster 2 was chosen as the optimal cluster. Optimal cluster is the biggest cluster containing more clients. The priority is regarded for the residents of Sanandaj city, the disabled, females, uneducated, the married, and the number of people supported by the State Welfare Organization=1. Furthermore, in order to obtain the association between attributes, Chi-square test was applied. We find that all of them have pairwise dependency (p<0.05) except gender and educational degree, the number of persons supported by The State Welfare Organization and place of living (city/village) attribute, the educational degree and the place of living (city/village) attribute. Conclusion: According to the information obtained, The State Welfare Organization should pay more attention to the optimal cluster’s users. In the other words, it should focus on the clients living in Sanansaj, the disabled, the females, the uneducated, the married couples, the number of persons supported by the State Welfare Organization=1. Furthermore, after the implementation of clustering method, new State Welfare Organization of Iran’s clients can join the clusters with their attributes and it can help the State Welfare Organization to analyze their needs. Thus, due to the existing relationships between attributes, providing facilities based on one attribute can improve welfare based on the other attributes.