Use of Machine Learning in Traffic Prediction Using Deep Belief Network and Random Forest

Abstract

The prediction of traffic is very much essential to reduce different types of complications of roads. The flow of traffic depends on the time, place, weather conditions, and road condition concern zone. The flow of traffic basically depends on the real dataset of a specific area. This dataset comes from various types of sensors and cameras that are implemented in a specific area. Nowadays the concept of deep learning has accrued all the attention of traffic flow detection. In this research, the “Deep Belief Network” and Random Forest are applied with the online dataset of a specific area. The flow of traffic prediction is based on the dataset that is taken from the online source. There are different types of attributes are taken place in the flow of traffic prediction. These attributes are zone name, weather, temperature, and times are used to measure the flow of traffic in a specific zone. The first algorithm is proposed the DBN method that is used in the machine learning process to generate unsupervised learning. The second algorithm, RF is used to appropriate output result of the flow of traffic prediction. This algorithm is proposed “threshold-based techniques” for the accuracy and the improvement of the

output result of the prediction of traffic. The performance of the process of machine learning evaluated the precision and the accuracy. The DBN algorithm is very much efficient and it can give appropriate output results of traffic flow. The process of machine learning is used to predict the flow of traffic, so this project is very much successful.