Adaptive range in FIMT-DD tree for large data streams

Abstract

The number of vehicles that exists on public roads have increased drastically over the years. This have caused several problems, where one of the most common problem is traffic jam. There have been several studies that have tried to solve this problem, such as by using real time videos with computer vision, wireless sensor networks, and traffic data predictions. In this study, we proposed a modification of Fast Incremental Model Trees with Drift Detections (FIMT-DD) to predict the traffic flow from a large traffic data set provided by the Government of United Kingdom. From our experiment results using large datasets, our proposed method have proven to be more accurate in predicting the traffic flow as compared to the conventional FIMT-DD Algorithm.