GEH Statistic

The GEH statistic, developed by Geoffrey E. Havers in the 1970s during his tenure as a transport planner in London, England, serves as a critical tool in the field of traffic engineering, forecasting, and modeling. Its primary function is to provide a standardized method for comparing two sets of traffic volumes, typically those generated by a traffic model against actual observed counts.

While its mathematical structure bears resemblance to a chi-squared test, the GEH is fundamentally an empirical formula. This means it's derived from practical observations and has been proven effective through repeated application, rather than being a strict statistical hypothesis test. Its utility lies in its ability to offer a consistent metric for assessing the accuracy of traffic predictions across a wide spectrum of traffic volumes, a challenge that simple percentage comparisons often fail to address effectively.

A significant challenge in traffic analysis is the vast disparity in traffic volumes encountered across different road types. For instance, comparing the hourly flow on a major motorway (potentially 5,000 vehicles per hour) with that of a minor on-ramp (perhaps only 50 vehicles per hour) using simple percentages presents an inherent problem. A 10% error margin, acceptable for the motorway (500 vehicles), would be a substantial 50 vehicles for the on-ramp, a much larger proportion of its total flow.

The GEH statistic elegantly overcomes this by employing a non-linear relationship. This non-linearity ensures that a single, consistent threshold for acceptable deviation can be applied across a broad range of traffic volumes. This uniformity simplifies the validation process for traffic models, allowing engineers to maintain a consistent standard of accuracy regardless of the specific location or road type being analyzed.

The GEH statistic is calculated using the following formula: GEH = sqrt(2 * (M - C)^2 / (M + C)), where 'M' represents the hourly traffic volume from a traffic model (or a new count) and 'C' represents the real-world hourly traffic count (or an old count). The formula quantifies the difference between the modeled and observed volumes relative to their sum. A GEH value of 0 indicates a perfect match between the model and reality.

As the difference between M and C increases, the GEH value rises. In practice, traffic modeling guidelines, such as those from the UK Highways Agency's Design Manual for Roads and Bridges (DMRB) and Transport for London, establish thresholds for interpretation. A GEH score below 5.0 is generally considered a good match, indicating that the model is performing well.

Scores between 5.0 and 10.0 warrant further investigation, suggesting potential areas for model refinement or data correction. A GEH score exceeding 10.0 strongly implies a significant problem, pointing towards either a flawed travel demand model or inaccuracies in the input data, such as data entry errors or fundamental calibration issues.

The GEH statistic is not merely an academic curiosity; it is a cornerstone of practical traffic analysis and is widely recognized and mandated in professional guidelines. Its adoption by major transportation authorities, including the UK Highways Agency, Wisconsin's microsimulation modeling guidelines, and Transport for London, underscores its importance. These bodies rely on the GEH to ensure the reliability of traffic models used for planning new infrastructure, assessing the impact of policy changes, and optimizing existing transportation networks.

For instance, in baseline scenario modeling, the DMRB specifies that 85% of traffic volumes should have a GEH less than 5.0. This rigorous application helps ensure that decisions about billions of dollars in infrastructure investment are based on sound, validated predictions, ultimately contributing to more efficient, safer, and sustainable transportation systems.

While the standard application of the GEH statistic focuses on hourly traffic volumes, its underlying principle can be adapted. However, it is crucial to convert traffic flows of different durations (e.g., 15-minute counts) into hourly equivalents before applying the standard GEH thresholds. This ensures consistency in interpretation.

The GEH statistic is closely related to the broader field of traffic model validation, which involves a suite of techniques to ensure that models accurately represent real-world traffic behavior. Other related concepts include sensitivity analysis, where the impact of changing model parameters is assessed, and calibration, the process of adjusting model parameters to best fit observed data. The GEH statistic acts as a key performance indicator within this larger validation framework, providing a quantifiable measure of model accuracy that directly informs the confidence engineers can place in their forecasts and planning outputs.