Analysis and Prediction of Motor Vehicle Carbon Dioxide Emissions Using a Hybrid LSTM and Arima Algorithm

CO2 emissions from motor vehicles contribute substantially to climate change. Accurate prediction of emission trends is thus crucial for mitigation strategies. This research evaluates the performance of a Hybrid Long Short-Term Memory (LSTM) and Autoregressive Integrated Moving Average (ARIMA) model for predicting Motor Vehicle CO2 Emissions. This hybrid model integrates ARIMA's capability in handling linear patterns and LSTM's in capturing long-term non-linear dependencies. Using 1000 historical data entries from the Eco-Route Application, the hybrid model was tested and compared with single models. Results show the hybrid model achieved good prediction accuracy with MAE 0.0941, MAPE 10.20%, and RMSE 0.1081 in its best scenario. However, on this specific dataset, the single ARIMA model demonstrated the best overall performance (MAE 0.0835, MAPE 9.33%, RMSE 0.0975). Dataset limitations were identified as affecting the hybrid's capability. The Hybrid LSTM-ARIMA model is determined to be a promising option for CO2 emission prediction, especially when larger datasets are available.