Scholarlink

<p>Accurate travel time prediction is essential for evaluating and improving the performance of transportation networks in Intelligent Transportation Systems (ITS). However, this task remains challenging due to dynamic traffic conditions and incomplete trip datasets, which often lack detailed trajectory information. To address these limitations, we propose a framework which includes a route reconstruction approach that maps origin–destination (O–D) pairs to an OpenStreetMap (OSM)–derived road network, enabling link-level travel time estimation. Our method integrates Uber Movement speed data, intersection delays, as well as external factors such as weather conditions, temporal patterns, and holidays into a comprehensive feature set. We evaluate our approach using XGBoost. The framework is validated on large-scale taxi trip datasets from New York City, Santiago, Porto, and Bangkok, demonstrating adaptability to varied urban traffic conditions. Experimental results show that the proposed approach outperforms state-of-the-art methods, achieving improvements ranging from 1.69% to 8.76% for NYC and 1.46% to 2.84% for Santiago.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Travel Time Estimation for Vehicles Using Link-Level Data Through Tree-Based Models

  • 期刊论文

  • 发表日期：

    2025年11月29日

查看全文
  • Nishtha Srivastava,
  • Bhavesh N. Gohil,
  • Suprio Ray

摘要

Accurate travel time prediction is essential for evaluating and improving the performance of transportation networks in Intelligent Transportation Systems (ITS). However, this task remains challenging due to dynamic traffic conditions and incomplete trip datasets, which often lack detailed trajectory information. To address these limitations, we propose a framework which includes a route reconstruction approach that maps origin–destination (O–D) pairs to an OpenStreetMap (OSM)–derived road network, enabling link-level travel time estimation. Our method integrates Uber Movement speed data, intersection delays, as well as external factors such as weather conditions, temporal patterns, and holidays into a comprehensive feature set. We evaluate our approach using XGBoost. The framework is validated on large-scale taxi trip datasets from New York City, Santiago, Porto, and Bangkok, demonstrating adaptability to varied urban traffic conditions. Experimental results show that the proposed approach outperforms state-of-the-art methods, achieving improvements ranging from 1.69% to 8.76% for NYC and 1.46% to 2.84% for Santiago.