基于recurrent neural networks的网约车供需预测方法

以网约车订单等真实数据为数据源，结合TensorFlow深度学习框架，利用循环神经网络（recurrent neural networks）方法，预测网约车在未来某时间某地点的订单需求量。提出改进LSTM RNN（长短时记忆循环神经网络）模型，经过对其优化和训练，能够有效预测网约车未来某时间某地点的供需量。对数据源进行可视化分析，排除不相关数据源干扰，以此为基础设计仿真实验。仿真实验表明，该模型的正确率比反向传播神经网络（BPNN）、回归决策树（DTR）、非线性回归支持向量机（SVR）以及随机漫步（RW）等模型高，同时，对长短间隔不同的历史数据有较好的记忆能力，在测试数据上有较强的泛化能力。

Prediction method of supply and demand for online car based on recurrent neural networks

Orders from online car as data sources, using TensorFlow and recurrent neural networks, to predict the supply and demand for online car at a certain point in the future. This paper presents the model of LSTM RNN, which is optimized and trained to effectively predict the supply and demand of the online car at a certain point in the future. Visual analysis of data source, help excluding uncorrelated data source, which is the basic to design simulation experiment. Simulation experiments show that the accuracy of the model proposed is higher than back propagation neural network (BPNN) and decision tree regression(DTR) , nonlinear support vector regression machine (SVR) and random walk (RW) , at the same time, the excellent memory capability of different length of historical data, and the excellent generalization capability on the test set.