Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

LibCity: A Unified Library Towards Efficient and Comprehensive Urban Spatial-Temporal Prediction

Jiawei Jiang, Chengkai Han, Wenjun Jiang, Wayne Xin Zhao, Jingyuan Wang

TL;DR

LibCity tackles reproducibility and fair comparison in urban spatial-temporal prediction by delivering a PyTorch-based, modular library with five modules (Data, Model, Evaluation, Execution, Configuration). It introduces atomic files as a unified data storage format and ships 55 datasets with 65 reproduced models across 9 tasks, complemented by automatic hyper-parameter tuning and a visualization platform. The work establishes a comprehensive model leaderboard and standardized evaluation protocols to accelerate research while enabling easy extension with new data and models. Overall, LibCity provides a standardized, scalable ecosystem that can unify benchmarking and drive rapid advancement in urban spatial-temporal modeling.

Abstract

As deep learning technology advances and more urban spatial-temporal data accumulates, an increasing number of deep learning models are being proposed to solve urban spatial-temporal prediction problems. However, there are limitations in the existing field, including open-source data being in various formats and difficult to use, few papers making their code and data openly available, and open-source models often using different frameworks and platforms, making comparisons challenging. A standardized framework is urgently needed to implement and evaluate these methods. To address these issues, we propose LibCity, an open-source library that offers researchers a credible experimental tool and a convenient development framework. In this library, we have reproduced 65 spatial-temporal prediction models and collected 55 spatial-temporal datasets, allowing researchers to conduct comprehensive experiments conveniently. By enabling fair model comparisons, designing a unified data storage format, and simplifying the process of developing new models, LibCity is poised to make significant contributions to the spatial-temporal prediction field.

LibCity: A Unified Library Towards Efficient and Comprehensive Urban Spatial-Temporal Prediction

TL;DR

LibCity tackles reproducibility and fair comparison in urban spatial-temporal prediction by delivering a PyTorch-based, modular library with five modules (Data, Model, Evaluation, Execution, Configuration). It introduces atomic files as a unified data storage format and ships 55 datasets with 65 reproduced models across 9 tasks, complemented by automatic hyper-parameter tuning and a visualization platform. The work establishes a comprehensive model leaderboard and standardized evaluation protocols to accelerate research while enabling easy extension with new data and models. Overall, LibCity provides a standardized, scalable ecosystem that can unify benchmarking and drive rapid advancement in urban spatial-temporal modeling.

Abstract

As deep learning technology advances and more urban spatial-temporal data accumulates, an increasing number of deep learning models are being proposed to solve urban spatial-temporal prediction problems. However, there are limitations in the existing field, including open-source data being in various formats and difficult to use, few papers making their code and data openly available, and open-source models often using different frameworks and platforms, making comparisons challenging. A standardized framework is urgently needed to implement and evaluate these methods. To address these issues, we propose LibCity, an open-source library that offers researchers a credible experimental tool and a convenient development framework. In this library, we have reproduced 65 spatial-temporal prediction models and collected 55 spatial-temporal datasets, allowing researchers to conduct comprehensive experiments conveniently. By enabling fair model comparisons, designing a unified data storage format, and simplifying the process of developing new models, LibCity is poised to make significant contributions to the spatial-temporal prediction field.
Paper Structure (35 sections, 16 equations, 3 figures, 6 tables)

This paper contains 35 sections, 16 equations, 3 figures, 6 tables.

Table of Contents

  1. Introduction
  2. Introduction
  3. RELATED WORK
  4. Urban Spatial-temporal Prediction
  5. Related Libraries
  6. Spatial-Temporal Data
  7. Urban Spatial-Temporal Data
  8. Atomic files
  9. Spatial-Temporal Prediction Tasks
  10. Macro Group Prediction Tasks
  11. Micro Individual Prediction Tasks
  12. Fundamental tasks
  13. The Library: LibCity
  14. Data Module
  15. Atomic Files
  16. ...and 20 more sections

Figures (3)

  • Figure 1: Overview of the LibCity library
  • Figure 2: The data processing flow in LibCity
  • Figure 3: Experiment Management and Visualization Platform