Machine Learning for Urban Traffic Carbon Emission Prediction in Norway: A Case Study of Bergen

Abstract

Norway’s goal of achieving carbon neutrality by 2050 requires precise monitoring and prediction of urban traffic emissions, a major contributor to national greenhouse gas output. This study applies three machine learning models—Random Forest Regressor (RFR), Gradient Boosting Regressor (GBR), and Artificial Neural Network (ANN)—to predict hourly urban traffic carbon emissions in Bergen, using a 2022–2023 dataset integrating traffic flow, vehicle type (electric vs. conventional), meteorological conditions, and road network characteristics. Results show GBR outperforms other models: it achieves a Mean Absolute Error (MAE) of 0.28 kgCO/h, Root Mean Squared Error (RMSE) of 0.39 kgCO/h, and Mean Absolute Percentage Error (MAPE) of 5.23%. Compared to RFR and ANN, GBR reduces MAE by 18.8% and 24.3%, respectively. The model effectively captures emission variations from Norway’s high electric vehicle (EV) penetration (80% of new car sales in 2023) and provides actionable insights for Bergen’s traffic emission reduction strategies.

Previous article in issue Next article in issue