Big Data Based Traffic Data Discovery Analysis Using Machine Learning

In modern transportation systems, daily advancements often result in traffic congestion due to the increasing influx of data within big data processing. However, many current systems fail to address high traffic data volumes adequately, overlooking the impact of dimension reduction on the overall load. Consequently, the high intensity of data exacerbates feature interdependencies, leading to inaccuracies in predictions. To address this challenge, we present a solution called the Adaptive Traffic Forecasting for Congestion (ATFC) system, which utilizes the scheduled time variance feature evaluation (STVFE) technique to enhance real-time transportation services. The initial phase involves preprocessing, where we assess the dataset's dimensions and gauge the Heavy Congestion Succeeding Rate (HCSR) by tracking vehicle transmissions across crossover lanes. Leveraging the HCSR rate, we determine frequency level disparities through the application of the Spider competence assessment (SCA) method. This enables us to reduce irrelevant variable features via the STVFE process. The chosen distinctive features are then subjected to spider optimization and trained using the Modified Feed-Forward Support Vector Machine (MFFSVM) Algorithm. This combination facilitates the prediction of traffic flow in terms of both variance levels and classification, resulting in enhanced precision and recall rates when compared to alternative systems.