Kümelemenin Apriori veri madenciliği algoritmasına etkisinin incelenmesi

Mobile computing and communication devices are widely utilized by users from different occupations and their usage is steadily increasing. Mobile communication characterizes our current information era. This rapid diffusion has had a direct effect on the applications proposed for smart devices in recent years. Many organizations collect and store data about their customers, suppliers and business partners. However, much of the useful marketing insights are hidden in that enormous amount of data. Data mining is the process of searching and analyzing data in order to find potentially useful information. In this study, as a preliminary approach, we have proposed an ontology-based methodology. Although data mining consists of a broad family of computational methods and algorithms, for this study, we have chosen the Apriori algorithm in order to determine association rules. Next, we have aimed at examining the effectiveness of different clustering algorithms when determining the association rules. Hence, we have compared the results of five different approaches. Three clustering algorithms are used: K-means, Expectation Maximization and Hierarchical Clustering. Most predictive association rules with best values are obtained by `K-means? and `Hierarchy-based? data mining methodologies. Therefore, we have proposed a new algorithm that combines these two algorithms and we have called it the `Hierarchical K-Means algorithm. The data analysis framework is applied to the data of mobile operating systems? users. By extracting most important information from consumer data, we claim that this framework may direct providers/application developers offer the right product/ advertisement to the right consumer.