Clustering algorithms resume the datasets into few number of data points such as centroids or medoids, which explain the entire dataset briefly. In the domain of data-driven machine learning, the more precision with the clustering rule leads directly to more precise classification, prediction, and recognition. We propose an efficient clustering method, which applies the paradigms - mainly 3D Gaussian model - to estimate the optimum cluster number, cluster border, and congestion coordinates to model the datasets of the natural distributions. This approach considers both qualitative and quantitative features of the dataset and calculates the best scale to analyze it. We used fuzzy logic to compare the models with data, to generate and rank the hypotheses, and finally to reject or accept the assumptions. The proposed approach which is called Fuzzy Gaussian Paradigmatic Clustering (FGPC) algorithm is used as the basis of a fast (with the complexity order of O(n)) and robust algorithm for identifying fuzzy models.