Add Spatial joins and allocation: the matching functions as joins, and the road from a database to an allocation model
A new page that connects the functions which match two domain units the way a JOIN does in SQL: join_equal_values and rlookup on a given key; the pair-producing spatial pre-selections (join_near_values, connect_matrix / dist_matrix, point_in_all_polygons, geos_overlay_polygon, impedance_matrix) as inner joins; connect, connect_info, point_in_polygon, impedance_table, capacitated_connect and discrete_alloc as left joins that already chose one match per row; and how a pair domain aggregates into a relation but not back. The last section traces, in five steps, how a customer-order-product database differs from a spatial interaction or allocation model: the join table is recorded there, and derived, chosen, distributed and finally optimised under competing claims here.
Spatial-join now opens with a sentence (it opened with a code block, which loses the search snippet) and names itself as the Cartesian-product form. Reverse links on How-To-Model, Relational-model-versus-Semantic-arrays, the Relational, Network, Allocation and point-polygon hubs, and the see-also of join_equal_values, join_near_values, connect_matrix, geos_overlay_polygon and impedance_matrix.
The page is not in the geodms.nl menu until nav/geodms.md in GeoDmsWikiToStaticHtml lists it.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
connect_matrix / dist_matrix: all points within a cutoff per feature (#1236)
New pages for the sparse-matrix variants of connect_info / dist_info, and links
from Network functions, Connect_info and dist_info. Records the pair-domain
result, that both cutoffs are required and combine as a minimum, that
max_nr_matches keeps the k nearest, the order of the rows, that a multi-geometry
is not split, and the _eq / _ne null-key rule.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>