Fuzzification of 2-D density based clusters using spline regression models

A new technique to perform fuzzification of Density based clusters of 2-dimensional data using regression models has been proposed here. Generally, for fuzzification in partition based clusters, one would compute the center of clusters and then assign memberships for the instances based on their relative distances from centers of all clusters, but the same approach cannot be used for density-based clusters as they attain arbitrary shape. So as to fuzzify these type of clusters on 2-D data, we instead compute the fitting curve for each cluster and assign membership for the points based on the relative distances between points and fitting curves of all clusters. B-spline and cubic spline regression models for open data cluster and least squares fitting of ellipses for closed data cluster were used to produce fitting curves. By doing so the new technique is capable of fuzzifying density-based clusters.