Data mining in public services: A research on municipality help desk data

Management information systems used by the goverments and public agencies are crucial in terms of acquiring latent information patterns comprised from big data generated by the developing smart technologies and provision of decision supports on future decisions to policy makers and public managers. Technology-based studies in public administration are generally conducted on the basis of theoretical and practical dimensions of "e-government". Data mining applications are usually studied with focusing on private sector data in disciplines such as management information systems, computer sciences and business administration. This study examines data mining, on the accounts of the disciplines of public administration and management information systems. In the empirical part of the study, fourth chapter, data mining process is implemented with public data, unlike the general tendency in the literature. The help desk data of the Kocaeli Metropolitan Municipality is used in the study. Preprocessing of data and classification methods are implemented via "Weka Machine Learning" tool. The help desk data is analyzed using a number of machine learning algorithms such as Naive Bayes, Support Vector Machine, K-Nearest Neighborhood and Decision Trees. The results were visualized with a business intelligence application called "Tableu". It was concluded that while there is an increasing awareness in reent years on big data technologies and data mining in governments and public agencies in Turkey, the number of applications and projects have still been outnumbered. In essence our study shows that, careful and accurate pre-processing of the raw data, qualitative or quantitative, has a direct impact on the accuracy of machine learning algortihms. Finally, it seems that big data and data mining applications can be effectively used by the public agencies to enhance government operations, to provide effective and efficient public services, and to improve the quality of public policy-making. Data mining is not only a software tool that contains numerical methods; but it includes methods and applications intented to solve the real world problems. This thesis is also important in that it shows data mining can be adopted in public services with a generic model.