Causality Enhanced Origin-Destination Flow Prediction in Data-Scarce Cities

TL;DR

CE-OFP tackles OD flow prediction in data-scarce cities by transferring urban knowledge from data-rich cities. It combines a RL-based causal graph search to learn universal feature relations, a causality-enhanced variational auto-encoder (CE-VAE) to reconstruct missing features with quantified uncertainty, and a graph-attention knowledge-distillation stage to transfer the data-rich OD predictor to data-scarce cities. The approach demonstrates that universal causal graphs exist across cities and that incorporating them into feature reconstruction and prediction improves cross-city transfer, achieving up to a 11% reduction in $RMSE$ over strong baselines. This work enables accurate OD flow forecasting in developing urban contexts and highlights the value of causal knowledge and uncertainty-aware reconstruction for transfer learning in urban analytics.

Abstract

Accurate origin-destination (OD) flow prediction is of great importance to developing cities, as it can contribute to optimize urban structures and layouts. However, with the common issues of missing regional features and lacking OD flow data, it is quite daunting to predict OD flow in developing cities. To address this challenge, we propose a novel Causality-Enhanced OD Flow Prediction (CE-OFP), a unified framework that aims to transfer urban knowledge between cities and achieve accuracy improvements in OD flow predictions across data-scarce cities. In specific, we propose a novel reinforcement learning model to discover universal causalities among urban features in data-rich cities and build corresponding causal graphs. Then, we further build Causality-Enhanced Variational Auto-Encoder (CE-VAE) to incorporate causal graphs for effective feature reconstruction in data-scarce cities. Finally, with the reconstructed features, we devise a knowledge distillation method with a graph attention network to migrate the OD prediction model from data-rich cities to data-scare cities. Extensive experiments on two pairs of real-world datasets validate that the proposed CE-OFP remarkably outperforms state-of-the-art baselines, which can reduce the RMSE of OD flow prediction for data-scarce cities by up to 11%.

Figures (11)

• Figure 1: Overview of CE-OFP.
• Figure 2: Encoder $q(Z,\hat{Y_e}|X)$ of CE-VAE.
• Figure 3: Decoder $p(X,\hat{Y_d}|Z)$ of CE-VAE.
• Figure 4: GAT based model for OD flow prediction via knowledge distillation.
• Figure 5: Sub causal graph sampled from the complete causal graph found in New York City. The partial results are convenient for checkinging the reliability of the causal graph from intuition.