SPI-LSTM approach for enhancing traffic flow prediction

The idea of traffic flow prediction for congestion management have been proposed in order to improve traffic management. Machine learning and deep learning algorithms have enabled this idea to grow, as it involves an excessive amount of traffic data and variables. The usage and analysis of these data or variables are essential but vary among studies hence producing different outputs and results. The different types of traffic data may affect the accuracy of congestion calculations. This study proposes a conceptual model known as the Speed Performance Index and Long Short-Term Memory (SPI-LSTM) Approach for Enhancing Traffic Flow Prediction in Smart Cities model based on speed variable analysis, which is common in most traffic datasets. This study has explored the potential of adapting analysed traffic data, by calculating the Speed Performance Index, into Long Short-Term memory machine learning and deep learning algorithms in order to perform traffic prediction. To test the hypothesis of whether traffic data influence prediction outcomes, an experiment was conducted using the Python programme to generate the expected and predicted outcomes as well as performing results analysis. The results were validated using evaluation metrics and then, compared with other existing models in order to analyse the performance of the proposed model. The validation and comparison results illustrated a positive performance when compared with existing models, hence, showing the potential of the proposed model to improve traffic prediction.