摘要：An essential aspect of the transport system is public passenger transport and the Public Transport (PT) movement prediction is significant issues faced in the transport planning area because of its operational importance. In recent years, Intelligent Transportation Systems (ITS) have received a growing amount of interest. There are many advances and innovative applications that have been introduced for a safer, highly efficient, and even congenial environment from PT. A reliable and efficient system of traffic flow prediction is required for accomplishing these applications that build an event with the application of ITS implementations to resolve the potential road situation in advance. However, the PT network efficiency plays the main role for all urban authority areas in which the advancement of both communication and location devices are randomly increasing the data availability generated over the operational platform. In order to recognize trends useful for improving the Schedule Plan, adequate Machine Learning (ML) approaches need to be implemented. Therefore, this paper focused in heterogeneous data that affect the prediction value which is utilized for predicting the demand transport required in the particular route and arrival time of public transport using Density-Based Spatial Clustering of Applications with Noise (DBSCAN) with Seasonal Autoregressive Integrated Moving Average (SARIMA) algorithm to analyze the forecasting of the real-time passenger demand dynamically endorsed the growth of the dynamic bus management and scheduling. Moreover, the accuracy of proposed SARIMA Model is compared with traditional hybrid model such as Gaussian Mixture Model (GMM) with ARIMA model for providing an efficient and robust prediction of PT based on passenger demand.
其他摘要：An essential aspect of the transport system is public passenger transport and the Public Transport (PT) movement prediction is significant issues faced in the transport planning area because of its operational importance. In recent years, Intelligent Transportation Systems (ITS) have received a growing amount of interest. There are many advances and innovative applications that have been introduced for a safer, highly efficient, and even congenial environment from PT. A reliable and efficient system of traffic flow prediction is required for accomplishing these applications that build an event with the application of ITS implementations to resolve the potential road situation in advance. However, the PT network efficiency plays the main role for all urban authority areas in which the advancement of both communication and location devices are randomly increasing the data availability generated over the operational platform. In order to recognize trends useful for improving the Schedule Plan, adequate Machine Learning (ML) approaches need to be implemented. Therefore, this paper focused in heterogeneous data that affect the prediction value which is utilized for predicting the demand transport required in the particular route and arrival time of public transport using Density-Based Spatial Clustering of Applications with Noise (DBSCAN) with Seasonal Autoregressive Integrated Moving Average (SARIMA) algorithm to analyze the forecasting of the real-time passenger demand dynamically endorsed the growth of the dynamic bus management and scheduling. Moreover, the accuracy of proposed SARIMA Model is compared with traditional hybrid model such as Gaussian Mixture Model (GMM) with ARIMA model for providing an efficient and robust prediction of PT based on passenger demand.
