Scalable Transit Delay Prediction at City Scale: A Systematic Approach with Multi-Resolution Feature Engineering and Deep Learning

2026-01-26

AI Summary

Urban bus systems lack scalable and reusable network-wide delay prediction capabilities, hindering real-time information services and scheduling optimization. This work proposes an efficient city-scale bus delay prediction framework that mitigates the “giant cluster” problem in dense urban areas by integrating H3 geospatial grids with topological clustering to form balanced route clusters. It introduces a multi-resolution (1,683-dimensional) cluster-aware feature representation, combined with adaptive PCA for dimensionality reduction and an LSTM model. Evaluated on June data from Montreal’s STM network, the approach achieves 18%–52% higher prediction accuracy than Transformer-based models while using only 1/275 of the parameters, enabling real-time, scalable deployment across entire urban transit networks.

Technology Category

• Planning, Routing, and Scheduling: Optimization of Spatio-temporal Systems
• Machine Learning: Scalability of ML Systems
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• Web Mining and Content Analysis: Large pretrained models with web data
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Urban bus transit agencies need reliable, network-wide delay predictions to provide accurate arrival information to passengers and support real-time operational control. Accurate predictions help passengers plan their trips, reduce waiting time, and allow operations staff to adjust headways, dispatch extra vehicles, and manage disruptions. Although real-time feeds such as GTFS-Realtime (GTFS-RT) are now widely available, most existing delay prediction systems handle only a few routes, depend on hand-crafted features, and offer little guidance on how to design a scalable, reusable architecture. We present a city-scale prediction pipeline that combines multi-resolution feature engineering, dimensionality reduction, and deep learning. The framework generates 1,683 spatiotemporal features by exploring 23 aggregation combinations over H3 cells, routes, segments, and temporal patterns, and compresses them into 83 components using Adaptive PCA while preserving 95% of the variance. To avoid the"giant cluster"problem that occurs when dense urban areas fall into a single H3 region, we introduce a hybrid H3+topology clustering method that yields 12 balanced route clusters (coefficient of variation 0.608) and enables efficient distributed training. We compare five model architectures on six months of bus operations from the Soci\'et\'e de transport de Montr\'eal (STM) network in Montr\'eal. A global LSTM with cluster-aware features achieves the best trade-off between accuracy and efficiency, outperforming transformer models by 18 to 52% while using 275 times fewer parameters. We also report multi-level evaluation at the elementary segment, segment, and trip level with walk-forward validation and latency analysis, showing that the proposed pipeline is suitable for real-time, city-scale deployment and can be reused for other networks with limited adaptation.

Problem

• transit delay prediction
• city-scale
• bus operations
• real-time prediction
• scalable architecture

Innovation

• multi-resolution feature engineering
• hybrid H3+topology clustering
• adaptive PCA
• city-scale delay prediction
• lightweight LSTM