Since intuitionistic fuzzy sets (IFSs) can effectively deal with fuzzy and uncertain data, this paper proposes a fuzzy C-means clustering algorithm (FCM)-based on intuitionistic fuzzy sets for the inaccuracy in real clustering problems. Aiming at the problem that the traditional FCM algorithm is sensitive to the selection of the initial cluster center, the density region is divided, and the initial cluster center is selected in the high-density region to avoid the noise in the low-density region. The intuitionistic fuzzy entropy is introduced to calculate the feature weight of the data set, and the feature value is weighted, and the influence of the feature weight on the clustering result is considered. Finally, the specific steps of the improved algorithm are given, and the feasibility and superiority of the method are illustrated by typical examples.