People think of themselves as living in neighbourhoods, or places that are sufficiently close to be experienced on a day-to-day basis. Very often neighbourhood is the basis of spatial context, characterising the nature of a person’s surroundings. Neighbourhoods are often conceived as partitioning an urban space, such that every point lies in exactly one neighbourhood, but this may conflict with individual perceptions of neighbourhood, and by the expectation that neighbourhood extends in all directions around every individual’s location.

Figure 2‑3 shows three examples of possible neighbourhood definitions. In Figure 2‑3A the neighbourhood is defined as a circle centred on the house, extending equally in all directions. In Figure 2‑3B neighbourhood is equated with an existing zone, such as a census tract or precinct, reflecting the common strategy of using existing aggregated data to characterise a household’s surroundings. In Figure 2‑3C weights are applied to surroundings based on distance, allowing neighbourhood to be defined as a convolution (see further, Section 2.3.10) with weight decreasing as a simple function of distance.

In many geospatial analysis applications the term neighbourhood is used to refer to the set of zones or cells that are immediately adjacent to a selected zone or cell. For example, a raster model is generally defined using a regular grid of square cells. The set of 8 cells that surround each cell (omitting the grid’s edge cells) is often referred to as the ‘Moore’ neighbourhood  (see further, Section 8.2.2) or ‘Queen’s move’ neighbourhood (by analogy with the game of chess). Likewise the subset that consists of just the direct North, South, East and West neighbours is sometime called the ‘von Neumann’  (see further, Section 8.2.2) or ‘Rook’s move’ neighbourhood. Note that whilst such neighbourhoods apply to regular grids, irregular lattices (e.g. administrative areas) also have neighbourhood structures of that may be described in this or a similar manner.