Project information

  • Category: Machine Learning
  • Data: Origin-destination demand data
  • Project: Work of PhD student - Zahra
  • Methods: Data fusion, LSTM, GCN+LSTM
  • Technologies: Python, pandas, geopandas, folium, tensorflow
  • Conference/Journal: TRB Annual Meeting 2022
  • URL: Preparing for journal

Project details

This research addresses the challenges in predicting large-scale trip production, a crucial element in OD demand estimation. It proposes an efficient solution integrating Graph Convolutional Neural Network (GCN) and Long Short-Term Memory Network (LSTM), utilizing a nationwide graph to handle spatial heterogeneity. Unlike traditional models that require individual training for each spatial area, this method employs a single model, significantly reducing computation time. The study reveals the impact of spatial uncertainty on predictions, providing key insights for improving OD demand prediction models.