Using spatial weights matrix#
To capture patterns in urban form, we have to analyze morphometric
characters within a spatial context, taking into the account
neighbouring elements. A simple example could be a mean height within
100 metres (measured as a location-based character for each building).
momepy is using
libpysal spatial weights to capture the
relationship between elements, in this case, buildings, in the form of a
binary matrix (1 = neighbours, 0 = not neighbours). Spatial weights are
an essential part of
momepy and are used on many occasions and in
various forms. However, they should always be based on the unique ID of
element, not index (there are a few exceptions documented in API).
This section covers: