This research employed machine learning approaches to classify acceptable or non-acceptable particulate matter (PM_2.5) concentrations using a dataset that was obtained from the Nairobi expressway road corridor. The dataset contained air quality data, traffic volume, and meteorological data. The Boruta Algorithm (BA) was utilized in conjunction with the Random Forests (RF) classifier to select the most appropriate features from the dataset. The findings of the BA analysis indicated that humidity was the most influential factor in determining air quality. This was closely followed by the variables of ‘day_of_week’ and the volume of traffic bound for Nairobi. The temperature of the site was determined to have a lower significance. The comparison among different machine learning classifiers for the classification of acceptable and unacceptable PM_2.5 concentrations revealed that the Extreme Gradient Boosting (XGBoost) classifier displayed superior performance in terms of Sensitivity (0.774), Specificity (0.943), F1-Score (0.833), and AU-ROC (0.874). The Binary Logistic Regression (BLR) model demonstrated comparatively poorer performance in terms of Sensitivity (0.244), Specificity (0.614), F1-Score (0.455), and AU-ROC (0.508) when compared to other ML models. The prediction of PM_2.5 has the potential to provide valuable insights to transport policymakers in their deliberations on urban transport policy formulation.

Oversaturation of highways occurs due to their inadequate assessment and design. In this paper, we propose both a mathematical queuing model and a Discrete-Event Simulation (DES) framework based on Newell’s triangular flow-density relationship for the performance analysis of a multi-lane highway section. The proposed framework is a finite capacity queuing system, which captures an increase in the flow with the vehicle density up to the capacity of the section in an unsaturated condition and a decrease in the flow in the case of a saturated condition, depicting the actual traffic conditions on the highway section. First, the Birth–Death Process is used to build the mathematical queuing model (BDP), and the average number of vehicles (average queue length) and blocking probability on the highway section are estimated. Then, the accuracy of the mathematical queuing model is verified by the proposed DES framework. The “significance and effects” of different design factors are evaluated using the two-level full factorial design technique. The analysis of the experimental results reveals that the length of the highway section and the number of lanes are the most significant factors affecting the average queue length and blocking probability, while the jam density only has a significant effect on the average queue length and does not affect the blocking probability. In case of a two-way interaction, the combined effect of the “length-lanes” significantly affects the average queue length. In the end, a multiple-factor linear regression model is also developed for the prediction of the average number of vehicles on the highway section based on the design factors.