Determining the optimal number of seasonal adjustment factor groupings when estimating annual average daily traffic and investigating their characteristics

Abstract

Although cluster analysis is recommended by the US Traffic Monitoring Guide (TMG) to supplement the development of seasonal adjustment factor groupings (SAFGs), the relationships among SAFGs' characteristics remain undiscovered, while the determination of the optimal number of clusters is an ambiguous task exposed to great subjectivity. Statistical indicators provide a mathematical solution by removing engineering judgment without taking into consideration any guidelines or other criteria, necessary for transportation planners to generate ‘practical and sensible’ groupings. The method examined in this study aims to overcome the above weaknesses incorporating into the methodology a series of statistics, recommendations, and previous research findings. The investigation of the relationships among (1) the within-group variation, (2) the total number of sites, (3) the minimum number of stations within a cluster, (4) the optimal number of clusters, and (5) the geographical size of the groups constitutes the main objectives of this research. According to the results, the cluster variability declines as the available number of stations increases. When the minimum number of stations within a cluster increases, the weighted coefficient of variation inflates as well, with the rate of increase depending on sample size. The average number of automatic traffic recorders per cluster is analogous to the sample size, while the optimal number of clusters varies conversely with the minimum number of stations within a cluster. The application developed for the conduct of the analysis minimizes the computational time needed, while it can be easily implemented by engineers to automate the process recommended by the TMG, enhancing the current state of practice.

Keywords:

• annual average daily traffic
• seasonal adjustment factors
• automatic traffic recorders
• traffic monitoring program
• cluster analysis
• optimal number of clusters

Acknowledgments

The author of this paper would like to thank Dave Gardner, Tony Manch, and Lindsey Pflum from the Ohio Department of Transportation for their valuable advice. The contents of this paper reflect the views of the author, who is responsible for the facts of the data presented herein. The contents do not necessarily reflect the official views or polices of ODOT or FHWA.

Disclosure statement

No potential conflict of interest was reported by the authors.