Time-Series Forecasting of Trip Generation/Attraction using Deep Learning Model

Abstract

When traffic congestion forecasts and countermeasures them, it is necessary to estimate traffic demand properly. In this study, an estimate model of trip generation/attraction of the next day based on past traffic volume records, using deep learning models such as LSTM (long short-term memory). Specifically, using the data collected by ETC2.0 before and after the Tokyo Olympics and Paralympics 2020, time-series learning data is constructed and modeled.