A Stochastic Approach for Modeling the Saturation Flow Rate at Traffic Signals in Jordan

Evaluating the saturation flow rate (SFR) is crucial for determining infrastructure capacities, signal timing, and level of service (LOS) at signalized intersections. This study modeled saturation flow rate using two approaches: the classical frequentist model and the stochastic model developed using Bayesian and Cholesky decomposition. This study showed that turning radius, camera enforcement, speed limit, and movement type all greatly affect SFR. It did this by collecting detailed data from 19 signalized intersections in Jordan and calibrating the model. This showed that the stochastic model can handle the natural variability in traffic flow data. The results indicated that the four studied variables significantly affected the SFR estimation. The models showed that the SFR increases as it experiences a rise in the turning radius, the speed limit, and the presence of camera enforcement (i.e., positive model coefficients). At the same time, it decreases with the movement type (i.e., negative coefficient). Using the Cholesky decomposition method on Bayesian data to find model coefficient correlations was limited to twelve parameters (coefficient means, variances, and correlations), whereas monitoring the stochasticity present in the actual data. It was found that Cholesky decomposition and Bayesian statistics were both able and valid in their estimation of coefficients. They came up with almost the same results.