Guide on Methods for Assigning Counts to Adjustment Factor Groups (2024)
Chapter: 1 Introduction
CHAPTER 1
Introduction
Background
Annual average daily traffic (AADT) represents the mean traffic volume across all days of a year for a given location along a roadway (Federal Highway Administration [FHWA] 2018). AADT is widely used by many agencies to meet reporting requirements, allocate resources, inform decision-making, perform analyses, and support various agency functions related to planning, design, operations, safety, maintenance, environmental analysis, and more. According to Highway Performance Monitoring System (HPMS) requirements, state departments of transportation (DOTs) must report AADT every year to FHWA for all federal-aid highways and grade-separated interchange ramps (FHWA 2016, FHWA 2022). States are also required by the Highway Safety Improvement Program Final Rule to have access to AADT for all paved public roads, including non-federal-aid system roads, by the year 2026 (FHWA 2016).
Transportation agencies calculate AADT for a select number of roadway locations where continuous count stations (CCSs) have been permanently installed. The purpose of the CCSs is to collect traffic data 24 hours a day, seven days a week, for all days of the year. Due to the high installation, operation, and maintenance costs of CCSs, it is economically difficult to cover the entire transportation network with CCSs. For many roadway segments that are not being continuously monitored by CCSs, agencies typically deploy less expensive portable traffic recorders (PTRs) to conduct short-duration counts (SDCs), which for simplicity are often referred to as “counts.” The duration of SDCs usually ranges from a few hours to a few days, with the most common durations being 24 or 48 hours. The goal is to develop the most accurate AADT estimates possible by factoring (also known as annualizing, expanding, or adjusting) the SDCs.
Agencies estimate AADT using variations of a traditional approach that was first introduced in 1966 by Drusch (Drusch 1966). An improved version of this approach is recommended by FHWA’s Traffic Monitoring Guide (FHWA 2022). The traditional approach combines CCS and SDC traffic data and includes four general steps:
- Step 1—Computation of adjustment factors for each CCS. This step involves processing CCS data and calculating AADT and various sets of temporal adjustment factors (e.g., 12 monthly factors, 84 monthly day-of-week factors) separately for each CCS.
- Step 2—Establishment of adjustment factor groups. This step is known as the grouping process and involves creating groups of CCSs that exhibit similar traffic patterns. The goal is to produce internally homogeneous and well-defined groups with easily identifiable characteristics that allow direct assignment of every SDC to a group. After the factor groups are developed, group adjustment factors are separately computed for each group.
- Step 3—Assignment of SDCs to factor groups. This step involves assigning each SDC to one of the factor groups based on one or more assignment attributes. The assignment attributes may be different than the grouping attributes used in Step 2.
- Step 4—Annualization of SDCs. This step, widely known as factoring, involves applying one or more group adjustment factors to each SDC to produce an AADT estimate.
One caveat of the traditional AADT estimation approach is that the accuracy of the predictions is subject to errors at each step. Previous research has shown that the assignment process is the most critical part of the traditional AADT estimation method (Davis 1996, Sharma et al. 1996). It can affect the AADT accuracy to a larger degree than the duration of SDCs (Sharma et al. 1996) and is largely affected by errors stemming from engineering judgment (Ritchie 1986). Commonly used methods for estimating AADT do not adequately address how SDCs should be assigned to adjustment factor groups. Also, there are concerns about their applicability to roadways with insufficient CCS traffic data, as well as about the accuracy of the derived AADT estimates.
Scope
The scope of this guide is to provide transportation agencies with information about existing, improved, and new methods that can be used to assign short-duration traffic volume counts to adjustment factor groups to develop AADT estimates for all roadway functional classes (FCs) and traffic volume ranges. The guide also provides information on annualizing counts using segment-specific adjustment factors developed from probe data. The latter refer to timestamped location data collected from phones, tablets and mobile devices, global positioning system (GPS) devices, smartphone applications, connected and autonomous vehicles, and other technologies embedded in vehicles. This guide was developed under NCHRP Project 07-30.
Target Audience
The target agencies of the guide are state and federal agencies, metropolitan planning organizations, local governments, and other transportation agencies that conduct counts and develop or use AADT estimates. The main audience within the target agencies is traffic monitoring staff, such as managers, data analysts, database administrators, planners, data technicians, and information technology personnel, who typically collect, process, analyze, develop, integrate, archive, publish, or report AADT estimates. A secondary but larger audience includes those who do not work directly in traffic monitoring programs but use AADT in different agency applications and functions and may be interested in learning more about the AADT accuracy associated with different AADT estimation methods.
The guide aims to assist not only those who are experienced in AADT estimation but also those who are new to the traffic monitoring field. It includes methods suitable for agencies that do not have CCSs on lower functional classes, those interested in improving existing or developing new adjustment factor groups, and those that desire employing more advanced statistical and machine learning methods.
Organization of the Guide
The chapters and appendices of this guide are organized as follows. Chapter 2 presents various traffic monitoring terms, concepts, data types, and traffic parameters that readers need to be familiar with to better understand the methods described in the remaining chapters. Chapter 3 describes how hierarchical and non-hierarchical clustering can be used to create groups of CCSs, as well as how SDCs can be assigned to clusters in a data-driven manner using a machine-learning method called decision trees. The methods presented in this chapter apply only to higher functional classes where many CCSs typically exist. Chapter 4 describes traditional methods that can
be used to develop factor groups and assign counts to them. The methods presented in this chapter apply to all functional classes, including lower functional classes. Chapter 5 describes how probe data can be used to develop segment-specific adjustment factors and then annualize SDCs. Chapter 6 summarizes key findings from the research performed in NCHRP Project 07-30. Appendices A, B, and C describe relevant grouping and assignment processes used by the North Carolina Department of Transportation (NCDOT), the Washington State Department of Transportation (WSDOT), and the Wisconsin Department of Transportation (WisDOT), respectively. Appendix D provides an example of how to apply the k-prototypes clustering algorithm in the open-source programming language R.
