Predictive Analytics for Traffic Flow Forecasting Using Enhanced K-Nearest Neighbours Algorithm

Abstract

Traffic flow forecasting plays an important role in route guidance and traffic management. Traffic flow prediction is an important precondition to lessen traffic congestion in large-scale urban areas. k-Nearest Neighbour (KNN) is one of the most important methods in traffic flow forecasting, but some disadvantages prevent the widespread application. For traffic flow prediction, the proposed work is concentrated on reducing the time complexity as well as improving the accuracy of prediction. By using the clustering mechanism, the time complexity of the algorithm is reduced. By twofold clustering, the data to be analysed by the algorithm is segregated and hence the accuracy is improved. For improving the accuracy of prediction we use a multivariate approach. We also provide a route guidance with traffic flow, which adds novelty to the concept. To implement the concept, we use publicly available London traffic flow dataset. The concept can be further investigated by considering real-time traffic flow data.