Fuzzy spectral clustering for identification of rock discontinuity sets

A novel method for graphical representation of discontinuity sets is presented, in which the degree of membership of each discontinuity to its assigned set is explicitly indicated by means of a color scale. Comparison between the computed clustering partitions when different algorithms are employed to identify three discontinuity sets has been done. An approach for graphical representation of discontinuity to its assigned discontinuity set is explicitly indicated by means of a scale of shaded colors. The transformation in the point of transformed space are expected to cluster around k orthogonal points on the surface of the k-dimensional unit sphere. The results reveal that the tight clusters that observations are expected to form in the transformed space, the computed degrees of membership of discontinuity observations to discontinuity sets are higher in general for the fuzzy spectral clustering algorithm than for the fuzzy k-means algorithm.