Projection-Based Partitioning for Large, High-Dimensional Datasets