Improving Stratification Procedures and Accuracy of Annual Average Daily Traffic (AADT) Estimates for Non-Federal Aid-System (NFAS) Roads

Tsapakis, Ioannis; Das, Subasish; Anderson, Paul; Jessberger, Steven; Holik, William

doi:10.1177/03611981211043544

Cited by 3 publications

(6 citation statements)

References 14 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Tsapakis et al used probe-based annual average daily traffic (AADT) estimates provided by StreetLight and determined their accuracy by comparing them with multiple state and local agencies’ observed traffic volume data for two areas in Texas ( 2 ). Statistical analysis of the data showed that the MAPE was 50%, which was 15% lower than the same measure determined previously for the transportation network in Minnesota using StreetLight data ( 3 ).…”

Section: Literature Reviewmentioning

confidence: 99%

“…Research on crowdsourced transportation data has evaluated the accuracy of StreetLight data using the mean absolute percent error (MAPE) (2,3), Akaike information criterion (4), and root mean square error (RMSE) (4). For example, for auto and bicycle modes, some regions with higher vehicle volumes have been identified to provide more accurate results than those with lower volumes (2,4). To date, no independent third parties have evaluated the accuracy of bus and rail ridership data from StreetLight.…”

mentioning

confidence: 99%

See 1 more Smart Citation

Examining Transit Activity Data from StreetLight Using Ridership Data from Virginia Transit Agencies

Raida,

Ohlms,

Chen

2023

Transportation Research Record: Journal of the Transportation R

View full text Add to dashboard Cite

Researchers and planners require ridership data to study factors that influence people’s choice to use transit. However, the data can be challenging to obtain directly from transit agencies. Crowdsourced big data platforms such as StreetLight promise easily accessible ridership-related data in standard formats. It is important to assess the reliability of these data, particularly for transit agencies serving small- to medium-sized cities, which are less likely than agencies in large cities to have ridership data in standard formats. In this study, hourly ridership data from 2019 were collected from four bus transit agencies and one rail agency in Virginia and compared with StreetLight data. Comparisons for rail data were made on a station-to-station basis. Bus data comparisons were made at the city-limit level and at an aggregated-route level for each agency. In sum, StreetLight could not provide 2019 bus activity data for more than half of the localities in Virginia. Comparisons between transit agency and StreetLight data showed smaller root mean square errors when longer periods were analyzed (e.g., 4 versus 2 months). Although order of magnitude of ridership may indicate whether StreetLight can provide bus activity data, the former was not found to be correlated with the accuracy of the latter. Using data from StreetLight’s current algorithm might not be appropriate without verification against agency data, especially for agencies in small- to medium-sized cities.

show abstract

Section: Literature Reviewmentioning

confidence: 99%

mentioning

confidence: 99%

Examining Transit Activity Data from StreetLight Using Ridership Data from Virginia Transit Agencies

Raida,

Ohlms,

Chen

2023

Transportation Research Record: Journal of the Transportation R

View full text Add to dashboard Cite

show abstract

“…Past research on AADT estimation can be broadly classified into three categories: traffic count-based, non-traffic countbased, and travel demand models. Traffic counts are typically estimated using methods relying on data from continuous and short-duration traffic counters [44]. Agencies generally adopt stratified sampling procedures to estimate AADT at noncovered local road locations.…”

Section: Literature Reviewmentioning

confidence: 99%

“…Agencies generally adopt stratified sampling procedures to estimate AADT at noncovered local road locations. The stratification is generally based on one or more attributes like the functional class type (say, urban or rural local road) [44]. Agencies collect traffic volume data at selected locations in each stratum and consider that as a representative of all the roads within the stratum [46], [47], [48], [49].…”

Section: Literature Reviewmentioning

confidence: 99%

One-Dimensional Convolutional Neural Network Model for Local Road Annual Average Daily Traffic Estimation

Mathew,

Pulugurtha,

Bhure

et al. 2023

IEEE Access

View full text Add to dashboard Cite

The focus of this research is to develop a robust model for accurately estimating link-level annual average daily traffic (AADT) of all the local functionally classified roads. The capabilities of onedimensional convolutional neural network (1D-CNN), a deep learning architecture, and the domain knowledge pertaining to local road travel characteristics were combined to estimate local road AADT. The AADT based on traffic counts collected at 12,899 stations on local roads in North Carolina during 2014, 2015, and 2016 were considered for model training, testing, and validation. A total of eight existing state-ofthe-art statistical, geospatial, and selected other machine learning models were compared with the 1D-CNN model to estimate local road AADT. These include ordinary least square (OLS) regression, geographically weighted regression (GWR), ordinary kriging, natural neighbor (NN) interpolation, inverse distance weighting (IDW), backpropagation artificial neural network (BP-ANN), random forest (RF), and support vector machine (SVM). The model development and test results showed that the 1D-CNN model performed better than the other considered models. The architecture of the 1D-CNN model can learn the intricate patterns in the local road AADT. The outputs from the methodological framework proposed in this research help practitioners perform safety evaluation, planning and implementing infrastructure improvements, fund allocation and prioritization, air quality estimates, and meeting Highway Safety Improvement Program (HSIP) reporting requirements.INDEX TERMS Annual average daily traffic, AADT, convolutional neural network, deep learning, local road, low volume road, machine learning.

show abstract

“…Departments of transportation (DOTs), metropolitan planning organizations (MPOs) and other organizations and agencies use traffic volume as a key input for roadway safety, design, planning, traffic operations, and pavement maintenance. Traffic volumes, especially average annual daily traffic (AADT), are used to calculate vehicle miles traveled (VMT) which helps in allocating funds toward roadway maintenance and safety improvements of works ( 1 ). DOTs continually collect traffic counts using both permanent count stations (i.e., automatic traffic recorders or ATRs) and temporary short-term count stations.…”

mentioning

confidence: 99%

Estimating Average Daily Traffic on Low-Volume Roadways in Louisiana

Yeboah

Codjoe

Thapa

2022

Transportation Research Record: Journal of the Transportation R

View full text Add to dashboard Cite

Traffic volume is an important parameter that agencies use as a decisive factor, especially at the time of design, maintenance, and operation of roadways. Thus, its correct estimation is very essential. There are already a few established proprietary products available on the market that can predict traffic volumes. In addition, past studies have used several mathematical models to predict traffic volume. Relatively fewer studies have been conducted on predicting traffic volume on low-volume roadways. To bridge the gap on the prediction of traffic volume for low- and high-volume roadways, this study seeks to find practical, cost-effective, and progressive methods of estimating and classifying traffic on low-volume rural roadways. Across the state of Louisiana, 395 locations with low traffic volumes of less than 500 vehicles per day were selected. Census tract data was used to extract demographic and socioeconomic information for each location. Two prediction models—linear regression and random forest regression models—were developed to predict traffic volumes on these low-volume roadways. The results showed that the linear regression model had the highest predictive accuracy, with R-square of 0.979 and root mean square error of 70.26 compared with the RMSE of 110.23 for the random forest regression model. Both models found functional class, land use, number of lanes, population density, median age, median household income, and household density significantly affecting the traffic volume.

show abstract

Improving Stratification Procedures and Accuracy of Annual Average Daily Traffic (AADT) Estimates for Non-Federal Aid-System (NFAS) Roads

Cited by 3 publications

References 14 publications

Examining Transit Activity Data from StreetLight Using Ridership Data from Virginia Transit Agencies

Examining Transit Activity Data from StreetLight Using Ridership Data from Virginia Transit Agencies

One-Dimensional Convolutional Neural Network Model for Local Road Annual Average Daily Traffic Estimation

Estimating Average Daily Traffic on Low-Volume Roadways in Louisiana

Contact Info

Product

Resources

About