Can I read Prediction of special OD passenger flow based on transfer learning on EtoBox?

Prediction of special OD passenger flow based on transfer learning by Yu Wang; Zhengpu Zhang; Xiaodong Wang; Cheng Tao is a Physics and Astronomy article available to read on EtoBox.

What is Prediction of special OD passenger flow based on transfer learning about?

## Abstract The prediction of railway passenger flow has always been an important topic in the field of railway operation. Due to the limitation of the number of allocated tickets, some special OD(Origin to Destination) data are sometimes inaccurate. It will obviously affect the final prediction accuracy if it is directly used as a sample. However, it is very difficult to restore the biased data to the true value. In this paper, we choose case-based transfer learning to remove the unqualified samples from the target domain and replace them with the corresponding samples in the source domain and form the final training sample data set. At the same time, we choose the improved AdaBoost algorithm, the sample weight is adjusted by the error size, and the model is retrained to get the final model. The results show that the prediction accuracy of transfer learning based on improved AdaBoost algorithm is better than that of traditional ensemble learning, ARIMA model and multiple regression model. It also provides a new theoretical view for special OD passenger flow prediction in the future.

Who reads Prediction of special OD passenger flow based on transfer learning?

Author

Yu Wang; Zhengpu Zhang; Xiaodong Wang; Cheng Tao

Publisher

IOP Publishing

Published

2021

Language

EN

Field

Physics and Astronomy (Physical Sciences)