Nearest neighbour analysis

Overview

What the technique measures, the scale it produces, and what values near 0, near 1 and approaching 2.15 each say about a distribution of settlements.

What you need to know

Answer each one before you reveal it.

Questions from this subtopic

Each one with the model answer underneath.

1. 1

   What does nearest neighbour analysis measure, and what data does it need?

   Model answer

   It measures how the points of a distribution are spaced across an area, producing a single figure that describes whether the pattern is clustered, random or regular. It needs the locations of a set of points within a clearly defined area whose size is known - settlements in a region, shops of one type in a town, farms in a parish, erratics on a surface - together with the straight-line distance from each point to its own nearest neighbour.

2. 2

   What scale of values does nearest neighbour analysis produce?

   Model answer

   The index runs from 0 at one extreme to 2.15 at the other, with 1.0 in the middle. Zero represents complete clustering, 1.0 represents a random distribution, and 2.15 represents a perfectly regular distribution. Real-world results almost always fall somewhere between the extremes, so the interpretation is about which end of the scale the value leans towards.

3. 3

   What does a nearest neighbour value close to 0 indicate?

   Model answer

   It indicates clustering: the points are grouped tightly together, so the average distance from each point to its nearest neighbour is very small compared with what would be expected if they were scattered at random over the same area. Geographically, clustering points to a strong localising factor - settlements concentrated on a coalfield, at a bridging point, along a valley floor, or shops of one type grouping together to benefit from comparison shopping.

4. 4

   What does a nearest neighbour value close to 1 indicate?

   Model answer

   It indicates a random distribution: the points are neither noticeably grouped nor noticeably evenly spread, and the spacing is what chance alone would produce over that area. Geographically this suggests that no single strong factor is controlling location, or that several competing factors are cancelling each other out.

5. 5

   What does a nearest neighbour value approaching 2.15 indicate?

   Model answer

   It indicates a regular, evenly spaced distribution, the theoretical maximum being a perfect triangular lattice in which every point is the same distance from its neighbours. Geographically, near-regular spacing suggests competition for space or resources holding the points apart - market towns each needing their own trade area, farmsteads each needing their own land, or service centres spaced out so their catchments do not overlap, which is the pattern central place theory predicts.

6. 6

   An analysis of the distribution of villages returns a value of 0.4. How would you interpret and explain that?

   Model answer

   A value of 0.4 lies well below 1.0 and towards the clustered end of the scale, so the villages are significantly grouped rather than randomly or evenly spread. The interpretation is that some factor is drawing settlement into particular places - a valley floor with water supply, shelter and flatter land, a coastal strip, a former mining area, or the avoidance of high, wet or steep ground elsewhere in the study area. The figure describes the pattern; the geography must be supplied by relating it to the relief, drainage, geology and history of the area.

Other ways to work this

The same content, four other ways.

What is inside Levo

Six ways to work the same material.