Fig. 2
From: Managing distance and covariate information with point-based clustering
Panel (a) shows the small area meshblocks for urban Invercargill. The network represents the nearest neighbour connections used for spatial correlation. Clustering of New Zealand Deprivation Index is shown in panels (b) and (c). Moran’s I and the autocorrelation coefficient [19] are shown for increasing lag (steps) from any meshblock. The network model is used to determine the nearest neighbour (lag 1), 2nd nearest neighbour (lag 2), etc. Both measures show significant clustering of deprivation for several neighbourhood steps. The associated frequency of DSH index episodes and deprivation is shown in panel (d). The linear model (dashed line) has an adjusted R2 = 0.69