A Model for Predicting Traffic Congestion Using Deep Learning Algorithm: Case of Nairobi Metropolitan

Abstract

Traffic congestion is a widespread problem that plagues urban transportation systems, causing delays, increased fuel consumption, and environmental pollution. Addressing this issue requires accurate prediction of traffic congestion, enabling proactive management strategies and real-time information dissemination. Deep learning algorithms have emerged as powerful tools for traffic prediction, offering the potential to forecast congestion patterns effectively. The development of a model for predicting traffic congestion that is capable of accurately detecting and reducing the overall density of traffic in most urban areas frequented by motorists, such as offices, downtown, and establishments, has become one of the main challenges for engineers and designers in recent years. Traffic prediction models in use today are based on several modern technologies, including wireless sensor networks and surveillance cameras. In Kenya, the Nairobi Metropolitan Area has greatly felt the impacts of traffic congestion due to ever growing urban population. This is primarily because the number of vehicles has rapidly increased as compared to the infrastructure growth. This study presented a platform for addressing the traffic congestion through the establishment of Intelligent Traffic Management model using Deep Learning Algorithm. The study utilized observation checklist and questionnaire as the source of data for the study. An observation data collection sheet was used in collecting the data from the four main roads. To obtain data from the traffic officers, questionnaires was used. SPSS version 28 were used to analyze the data. Further from the correlation analysis, all the variables including High cost of travel/fares (r=.494), High vehicle maintenance (r=.206), Environmental pollution (r=.359), Staff fatigue (drivers and conductors) (r=.488), Accidents (r=.310), Poor road design (r=.308), Poor Traffic control system (r=.410), Road construction and maintenance works (r=.353), Vehicle break downs (r=.179), Roadside parking/obstruction (r=.452), High number of private cars (r=.233), High number of public transport vehicles (r=.071), Behavior of road usage (r=.228) Accidents (r=-.042), Poor road use (r=-.042) and Poor traffic management (r=-.209) had positive correlation with traffic congestion in Nairobi Metropolitan Area. Regression analysis further found that poor traffic management by traffic officers, a high number of public transport vehicles, poor road design, accidents, a high number of private cars, poor road use, poor traffic control systems, driver behavior, vehicle breakdowns, road construction and maintenance, and roadside parking explained up to 34.4% of the variation in travel time. In comparison, factors such as driver behavior, roundabout type, time of day, number of lanes, vehicle type, weather conditions, and travel rate explained 13.7% of the variation in road travel rates. Therefore, improved infrastructure, traffic management practices, and enhanced driver behavior are concluded to reduce travel time and improve transportation efficiency in the region. The study recommends that traffic engineering and urban planning practices should prioritize the optimization of road networks. The study recommends that local authorities and law enforcement agencies should collaborate to enforce traffic rules and regulations rigorously. The study also recommends that implementation of robust traffic management strategy by improving traffic signal synchronization, implementing intelligent traffic management systems, and investing in technology-driven solutions like real-time traffic monitoring and congestion alerts. Adequate and efficient traffic management by officers should also be ensured, as this factor has been found to play a substantial role in congestion mitigation. Additionally, policymakers should consider congestion pricing mechanisms during peak hours. This will incentivize drivers to use alternative routes or modes of transportation, thus reducing traffic congestion during high demand periods. Revenues generated from congestion pricing can be reinvested in transportation infrastructure and improvements.