Predicting Emission Costs for Urban Transportation in Smart Cities using Machine Learning Models

Singapore ratified the Paris Agreement on September 21, 2016, and the unconditional target by 2030 is the reduction of emission by 36%. Furthermore, the cities are the major source of emission, and transportation accounts for about 14% of the total emissions in Singapore. Transportation is one of the necessary infrastructures for cities and over 80% of the population in Singapore depends on public transportation systems. To significantly cut emissions, smart cities are expected to play an essential role, therefore the conventional transportation systems are transforming to be intelligent. The emission of vehicles depends on various factors like the average speed of the vehicle, total distance covered by the vehicle, urban driving conditions, and type of vehicle. In this research, COPERT model is used for calculation of emission cost after extracting features from real-world GPS datasets, and machine learning models are built to predict the total emission cost of the buses in an area of Singapore. In this study, the performance of emission cost estimation using three modeling techniques, namely MLP, RF, and Light GBM are compared. Results prove that the prediction accuracy of the proposed model is 97% when Light GBM is implemented.