An Improved Fuzzy C-Means Clustering Algorithm Considering Data Density Distribution

A refined fuzzy C-means clustering algorithm, utilizing a density-sensitive distance metric, is introduced. The traditional FCM algorithm is affected by noise and outliers when dealing with data sets with different density distributions, which leads to the deterioration of clustering performance. In order to overcome this problem, a density-sensitive distance metric is introduced, which can adaptively consider the local density information of data points, so as to reflect the similarity between data points more accurately in the clustering process. In the improved algorithm, instead of the traditional Euclidean distance, a density-sensitive distance metric is employed to determine the proximity of data points to cluster centers. This can effectively improve the robustness of the algorithm to noise and outliers, and better adapt to data sets of different densities. The experimental results show that compared with the traditional FCM algorithm, the FCM algorithm based on the improved density-sensitive distance significantly improves the clustering performance, especially when dealing with data sets with complex density distribution, showing better stability and accuracy.