A Traffic Jam Prediction Model for Dynamic Routes Using Global Positioning Systems Data from Vehicles at Different Times of the Day

摘要

Similar to most metropolitan cities like Tokyo, Beijing and Hong Jong, traffic Jams in Dhaka city, the capital of Bangladesh is very common due to dense population, lack of following traffic rules as well as inadequate pedestrian space. Traffic Jam Prediction is usually not precise because of the lack of data from the cars themselves. This paper aims to serve the purpose of predicting certain traffic-related information for a given route which was divided into a certain number of segments for ease of calculation. The amount of traffic-related information can be defined as the percentage of traffic jam for the 'whole route percentage' of traffic jam for each segments in the route, average time required to cover the whole route, average time required to cover each segments in the route, average speed you can have for the whole route and average speed you can have for each segments. Here, each segment is a small splice in a given route for simplicity that are integrated together in the end. All these results can be calculated for any given time of the day. The data-set for this research has been taken from a set of 700 cars on which Global Positioning System (GPS) Vehicle Tracking devices has been installed, continuously transmitting data of their state to the servers. We statistically show that unlike New York City, Dhaka sleeps at night - the proof being the density of the frequency of cars plummet at night.