The K-Means Clustering Architecture in the Multi-stage Data Mining Process

Abstract

In this paper, we used software engineering principles for the development of models and proposed the K-Means clustering architecture implemented on the multi-stage data mining process. We developed a modified architecture and expanded it by showing refinements on every process of the clustering and knowledge discovery stages. We used the mentioned hierarchical clustering model to partition the data into smaller groups of attributes so that we would determine the data structure before applying the data mining tools. The experiment shows that the model using the clustering resulted to an isolated but imperative association rules based on clustered data, which in return could be practically explained for decision making purposes. Shorter processing time had been observed in computing for smaller clusters implying faster and ideal processing period than dealing with the entire dataset.