On the data clustering algorithms based on density

The rapid development of technology has led to a large number of obtained and stored data. The rapid increase in data sizes caused the stored data to become meaningless and makes it difficult to reach meaningful information. In order to solve this problem, data mining has emerged. One of the most frequently used methods of data mining is the clustering analysis. Clustering analysis is a process in which, without prior knowledge, natural structures in the data set are discovered and relationships in the data set can be discovered. The most common problem in clustering analysis is the discoverability of arbitrary shaped clusters. Real world datasets can contain clusters in many different shapes and structures. A good clustering algorithm is expected to be able to discover clusters with arbitrary shape and structures. In order to solve this problem, density-based clustering algorithms have been developed. In this thesis study, clustering analysis, clustering methods and clustering approach based on density have been examined in detail. A new algorithm is proposed for the density-based clustering problem and two different versions of this algorithm are introduced. The introduced algorithms were run in a program prepared using the C # software language; the results obtained are visualized on this program. Calculation experiments performed through the prepared program were performed on nine actual data sets. According to the obtained results, it has been observed that the introduced algorithms can successfully obtain clusters in compact and well separated structures with arbitrary shapes. Key words: Data Mining, Clustering Analysis, Density-Based Clustering, C# Programming Language, Experimental Results.