The Prediction of Multimodal Public Transportation Sharing Rate Based on Data

Abstract

Accurate prediction of multimodal public transportation sharing rate is of great significance in coordinating traffic management, increasing public transport efficiency and allocating resources properly. The daily number of trips by subway, bus and ferry of pubic transport is calculated through data reduction and data mining, and the data of main factors affecting the fluctuation of public transportation sharing rate, i.e. holidays (or not), weather and air temperature, is collected in this paper based on big data on swiping public transportation IC cards in Shanghai. In addition, the sharing rates of subway, bus and ferry are predicted by using deep learning model based on historical data on daily number of trips and main influence factors, setting characteristic data and label data, and selecting activation function, loss function and gradient descent algorithm. The results show that the prediction error is less than 2.9%.