The New K-Means Initialization Method

Data clustering methods are crucial in various data analysis and machine learning fields. They are essential for organizing and understanding complex datasets by grouping similar (according to the assumed distance function) data. These methods facilitate customer segmentation, anomaly detection, pattern recognition and many other practical problems. Various clustering algorithms can be found in the literature, but the most popular method is the k-means algorithm and its subsequent modifications. This work presents a new method of initializing the k-means algorithm, which is tested on random and benchmark datasets. This paper discusses the importance of selecting the appropriate initialization of the k-means clustering method, demonstrating the effectiveness and superiority of the developed method.