Shared farthest neighbor approach to clustering of high dimensionality, low cardinality data