OPTICS is a state-of-the-art algorithm for visualizing hierarchical clustering structures of multidimensional objects. However, OPTICS requires iterative distance computations for all objects, and thus is computed in O(n²) time that makes it unsuitab...
OPTICS is a state-of-the-art algorithm for visualizing hierarchical clustering structures of multidimensional objects. However, OPTICS requires iterative distance computations for all objects, and thus is computed in O(n²) time that makes it unsuitable for massive datasets. Existing algorithms are not efficient to overcome the OPTICS issue. In this thesis, we propose a fast algorithm, called C-OPTICS, that alleviates the issue of time complexity using the constraint graph. C-OPTICS identifies the hierarchical clustering structures faster by reducing the number of distance computations while guaranteeing the hierarchical clustering structures identical to OPTICS. Both the qualities of the hierarchical clustering structures identified by C-OPTICS and its computational efficiency are demonstrated with experimental evaluations on synthetic and real datasets.