Scholarlink

<p>Trip pattern recognition, focusing on similarities in activity participation, is crucial for understanding and addressing individuals’ needs within the planning of a transportation system. This study presents a deep Natural Language Processing (NLP) clustering algorithm integrated with a multinomial logit model to extract and analyze individual activity-trip patterns from the 2017–2018 Regional Travel Survey (RTS) data from the Washington metropolitan area. Transforming daily activities of travelers into 288 five-minute intervals, the data was prepared as text and then fed into the NLP model, combined with a k-means clustering algorithm. This method identifies six distinct acitivity pattern clusters, encompassing full-time and part-time workers, retirees, students, and self-employed individuals. The model succeeds in recognizing distinct activity pattern clusters that were then interpreted looking at the relationships between activities, patterns, and socio-economic characteristics. It leverages the multinomial logit model to analyze the probability of cluster membership based on socio-economic factors, enriching the understanding of activity patterns across different demographic groups. This approach offers a nuanced perspective on urban populations' travel behaviors, facilitating targeted and equitable urban planning and policy-making. The study demonstrates the model's capability in handling complex activity sequence datasets, consisting of temporal activity sequences. This research underscores the value of integrating deep learning models in urban studies, opening new avenues for sophisticated, data-driven urban planning.</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

Activity-Travel Pattern Recognition of Individuals Using a Deep Short Text Clustering Algorithm

  • 期刊论文

  • 发表日期：

    2025年5月28日

查看全文
  • Hamid Mirzahossein,
  • Ali Bakhtiari,
  • Elham Shafidokht,
  • Sevgi Erdogan

摘要

Trip pattern recognition, focusing on similarities in activity participation, is crucial for understanding and addressing individuals’ needs within the planning of a transportation system. This study presents a deep Natural Language Processing (NLP) clustering algorithm integrated with a multinomial logit model to extract and analyze individual activity-trip patterns from the 2017–2018 Regional Travel Survey (RTS) data from the Washington metropolitan area. Transforming daily activities of travelers into 288 five-minute intervals, the data was prepared as text and then fed into the NLP model, combined with a k-means clustering algorithm. This method identifies six distinct acitivity pattern clusters, encompassing full-time and part-time workers, retirees, students, and self-employed individuals. The model succeeds in recognizing distinct activity pattern clusters that were then interpreted looking at the relationships between activities, patterns, and socio-economic characteristics. It leverages the multinomial logit model to analyze the probability of cluster membership based on socio-economic factors, enriching the understanding of activity patterns across different demographic groups. This approach offers a nuanced perspective on urban populations' travel behaviors, facilitating targeted and equitable urban planning and policy-making. The study demonstrates the model's capability in handling complex activity sequence datasets, consisting of temporal activity sequences. This research underscores the value of integrating deep learning models in urban studies, opening new avenues for sophisticated, data-driven urban planning.