Traffic Congestion Detection Using Machine Learning and Deep Learning

Abstract

Today, one of the nation’s biggest issues and one from which the people suffer greatly is traffic congestion. For individual passengers, business sectors, and governmental organizations, accurate and timely traffic flow information is today very important. It may aid drivers in making wiser travel choices, lessen traffic congestion, cut down on carbon emissions, and increase the effectiveness of traffic operations. To deliver such traffic flow data is the goal of traffic congestion detection. With the quick development and adoption of intelligent transportation systems (ITSs), traffic congestion detection is receiving more and more attention. It is considered to be a crucial component for the effective deployment of ITS subsystems, especially advanced traveler information systems, advanced traffic management systems, advanced public transportation systems, and advanced commercial vehicle operations. The study is about traffic congestion detection in intelligent transportation systems, which involves predicting between the previous year’s data set and the recent year data, which eventually delivers the accuracy and mean square error. Future construction of smart cities will be greatly reliant on intelligent traffic systems (ITS). And the innovation needed to integrate ITS into the infrastructure of smart cities is traffic congestion detection. This forecast will be useful for anyone who needs to check the current traffic situation. This traffic statistics is based on a one-hour time span. This forecast is used to assess live traffic facts. As a result, this will be easy to assess while the user is driving. The algorithm examines data from all routes to find the city’s most populous roads. After checking Bi-LSTM and other several models and calculation of RMSE, we finally offered a Trend/Linear Regression Model + Xgboost with Optuna for detecting traffic congestion using machine learning.