Expert Systems in Data Mining

Clustering is a fundamental task in data mining and knowledge discovery, playing a critical role in exploratory data analysis and expert system development. Among unsupervised learning techniques, K-Means clustering remains one of the most widely adopted algorithms due to its conceptual simplicity, scalability, and computational efficiency. Despite its extensive use, K-Means is often criticized for its sensitivity to initialization, outliers, and the requirement to predefine the number of clusters. This paper presents a conceptual and case-based analysis of K-Means clustering with a particular focus on its practical implementation in SPSS Clementine using real-world data. First, a conceptual framework is developed to clarify the theoretical foundations of clustering and the inherent ambiguity in cluster formation. Subsequently, a real-data case study is conducted to demonstrate the application of K-Means in an operational data mining environment, including data preprocessing, parameter configuration, cluster validation, and result interpretation using ANOVA and cluster-center analysis. The findings highlight both the strengths and limitations of K-Means when applied in expert systems and managerial decision-making contexts. The study concludes by identifying research gaps and future directions, emphasizing the need for adaptive and hybrid clustering approaches to enhance robustness and practical applicability in complex data-driven systems.

کلیدواژه ها:

Data Mining ، K-Means Clustering ، Unsupervised Learning ، Expert Systems ، SPSS Clementine