Congestion Aware Traffic Prediction System Based on Pipelined Time Variant Feature Selection for Improving Transportation of Real Time Service

Day to day development in transportation system the traffic congestion be occurred due to more data arrival in big data process leads more dimension. Most of the existing system doesn’t concentrates high traffic data volumes features by considering the burdens based on the dimension reduction. So, the intensive rate inspires the data non-alleviate for feature dependencies leads prediction inaccuracy. To resolve this problem, we propose a Congestion aware traffic prediction (CATP) system based on Pipelined Time Variant Feature Selection (PTVFS) for improving transportation of real time service. Initially the preprocessing was carried out verifies the dimension of the dataset and estimate the traffic intensive successive rate (TISR) by considering the vehicle transmission on crossover lanes. Based on the TISR rate the frequency level difference was estimated using the Spider fitness evaluation (SFE). This proposed system achieves high performance compared to the other system.