Spatial distribution of average predictability scores. (
a) Average predictability scores over London (in the 500×500
m grid pixels). (
b) Predictability scores at various aggregation levels as a function of the total activity volume. We see a clear trend with higher levels of activity resulting in higher average predictability scores; in the case of the pixels, the two measures have a Pearson’s correlation coefficient of 0.6121963 and a Spearman’s rank-correlation coefficient of 0.9032438, and can be fitted with the functional form
. Note that the fit was performed with taking only the pixels into account, but it also approximates the trend for the other spatial aggregation levels. The group of ‘outlier’ pixels (circled) can be identified as the vicinity of the Wembley
Stadium, a major
sports and event venue, where we can expect high volume of activity with less regularity. Map was generated with R, using the ggplot2, sp and rgdal libraries []; the background map was generated from the digital maps (shapefiles) available from the Greater London Authority []. Additional processing for the legend was done using Inkscape [].