compared to 2019 Georgia Department of Transportation (GDOT) traffic counts from the Traffic Analysis and Data Application (TADA) on a link-by-link basis for available roadways within the subarea. The 2019 counts were used to calibrate the travel demand model to pre-pandemic conditions. The deviation between the modeled and observed traffic was calculated for each individual roadway link where observed traffic count data was available. The traditional measures recommended by the Federal Highway Administration for model validation include the following: Coefficient of Determination (R2) and Root Mean Square Error (RMSE). These are used to provide a scalable measurement of model accuracy. The descriptions and results for each are outlined below. The correlation coefficient, known as R, is a standard statistical measure, calculated using the equation listed below. The coefficient of determination, known as R2, is the square of the correlation coefficient. R2 measures the overall model accuracy by showing how well the regression line represents the assigned data.
Ci = The observed traffic count for link i Vi = The modeled traffic volume for link i N = The number of links in the group of links including link i
FHWA suggests that an R2 result of at least 0.88 should serve as the standard for determining a model’s validity per TMIP Model Validation and Reasonableness Checking Manual in 2010. Figure C.2 below shows the variation of modeled volumes with the observed counts within the subarea. The R2 value is 0.9765, indicating that Streetlight’s Origin-Destination Data replicates the observed counts from GDOT’s TADA Counts from 2019.
F-4
Made with FlippingBook Ebook Creator