The GEH Measure and Quality of the Highway Assignment Models
O Feldman, Transport for London, UK
This note reports on some theoretical findings as well as analysis of different ATC data sets in order to get a better understanding of the nature of the GEH measure.
This paper reports on some theoretical findings as well as analysis of different ATC data sets so as to get a better understanding of the nature of the GEH measure. The goodness-of-fit measures suggested by the DRMB guidance about supplying assessments of transport model validity should take account of the variability of the count data. The GEH measure also confronts the significant issue of how to distinguish that a given level of absolute difference and of percentage difference can have very different levels of significance depending on the scale of the flows. Analysis of the different ATC data sets was carried out in order to understand variability of link flows, turning counts and enclosure flows. The research shows that the variability of traffic counts between comparable data on different survey days is significant and depends on road types, time of a day, area, and other factors. The variance-to-mean ratio can go up to 17, whilst the GEH values can go up to 67. In addition, the data suggest that some theoretical assumptions such as proportionality of variance to mean are quite strong assumptions and do not always hold.
The opinions expressed here are those of the author and not necessarily those of the Transport for London.
Association for European Transport