scipyspatial

scipy.spatial is a subpackage of SciPy that provides algorithms and data structures for geometric problems. It offers tools for distance computation, spatial indexing, and tessellations, and works with NumPy arrays across multiple dimensions. The module is designed to support scientific computing workflows that require spatial reasoning.

Distance computations are organized under scipy.spatial.distance. It includes functions such as pdist, cdist, and squareform, enabling pairwise distances within a collection and between two collections. A variety of distance metrics are supported, including Euclidean, Manhattan, and cosine, making it suitable for similarity search and clustering preparation.

Nearest-neighbor searches rely on KDTree and cKDTree, providing fast queries for nearest, radius-based, and range-limited neighbor searches. These structures support methods like query and query_ball_point to retrieve the closest points or all points within a radius, with performance benefits in moderate dimensions.

Spatial tessellations and hulls are provided via Voronoi, Delaunay, and ConvexHull classes. These constructs enable geometric partitioning and mesh generation in one or more dimensions. The implementations typically rely on the underlying Qhull library and are applicable to a wide range of dimensionalities, though performance degrades in very high dimensions.

scipy.spatial is a core part of SciPy and relies on NumPy. It is widely used in data analysis, geographic information systems, computer graphics, and machine learning pipelines for tasks such as distance-based retrieval, spatial clustering preparation, and geometric computations.