首页|Systematic Study on the Forecasting of Transit Passenger Flow Based on Machine Learning with Multi-Source Data
Systematic Study on the Forecasting of Transit Passenger Flow Based on Machine Learning with Multi-Source Data
扫码查看
点击上方二维码区域，可以放大扫码查看
原文链接
NETL
In this paper， multi-source data is used as the input of the model， with four machine learning algorithms: BP neural network， RBF neural network， support vector machines， and least squares support vector machines。 The four single models and six combination models are used to forecast the traffic passenger flow， respectively。 This research aims at forecasting traffic passenger flow systematically with different machine learning algorithms。 Comparing different models， we draw three new conclusions: (a) multi-source data is better than single-source data， which means that few variables doesn't have enough information for prediction; (b) combination model is better than single model， so the best model can be found and; (c) a fixed weight combination model is better than a variable weight combination model。 The conclusions of this study will help to exploit the potential of multi-source data for forecasting traffic passenger flow and improve the accuracy and efficiency of traffic planning。
Jiangsu Key Laboratory of Urban ITS, Jiangsu Province Collaborative Innovation Center of Modern Urban Traffic Technologies, School of Transportation, Southeast Univ., Nanjing, Jiangsu 210096, China
School of Transportation, Southeast Univ., Nanjing, Jiangsu 210096, China
COTA international conference of transportation professionals
editor(CN)
CICTP 2018: intelligence, connectivity, and mobility
NETL
