Enhanced Fuzzy C-Means Clustering for Improved Data Analysis with Application to the Higgs Boson Dataset

Fuzzy C-Means (FCM) clustering is a widely used algorithm for data partitioning, particularly when clusters exhibit overlap. However, its performance can be significantly impacted by sensitivity to initial conditions and susceptibility to noise. This paper proposes an Enhanced Fuzzy C-Means (EFCM) algorithm designed to mitigate these limitations. EFCM integrates a density-based initialization strategy using kernel density estimation, employs the robust Mahalanobis distance metric to handle noise and outliers, and incorporates Silhouette index-based adaptive parameter selection. The algorithm’s effectiveness is evaluated on the challenging Higgs Boson dataset from Kaggle, a high-dimensional and noisy dataset commonly used in high-energy physics research. Results demonstrate EFCM’s superior performance compared to traditional FCM in terms of clustering accuracy, execution time, and precision.

Keywords: Enhanced Fuzzy C-Means, Clustering, Fuzzy Clustering, Initialization, Noise Handling, Parameter Selection, Silhouette Index, Mahalanobis Distance, Higgs Boson Dataset, Kernel Density Estimation.