logo
CanyamAI summaries for academic research
language
MobileMobile
  • Phone FeatureWeChat Mini Program
EnglishEnglish
noticeNotice
BlogBlog
delete
arrow
Return

Spatiotemporal Changes in Traffic: A Comparison of Several Dimension-Reduction Methods Using a Railway Network Including Weather and Socioeconomic Variables

delete2026-03-01
delete0
PRE
AI
L
Loslever, Pierre *
DOI：10.1061/JTEPBS.TEENG-9095delete
deleteOriginal
deleteOriginal request for help
deleteShare
deleteSave
Abstract

Abstract

En 中文
The main aim of this paper is to compare dimensionality reduction methods for analyzing two databases: a database regarding traffic (DBT), and a database regarding the network environment (DBE), e.g., socioeconomic factors, pollution, weather, or accidents/incidents. In many cases, only one database is considered, whereas this paper suggests an analysis procedure where DBT is analyzed first for visualizing and interpreting complex traffic patterns, and DBE is investigated as extra data to explain these patterns. For DBT, the space unit is the origin-destination (OD) pair, and the time unit is a time window. Data are organized through a large table where the rows correspond to the time windows and the columns to OD pairs. From this large frequency table, several levels of time summarization are possible for showing data, like minutes, hours, or days; the same is true for space. The procedure uses the singular value decomposition principle with correspondence analysis (CA), taxicab correspondence analysis (TCA), and principal component analysis (PCA) as comparison methods. The comparison using an actual data set indicated that CA shows more interesting results. Results with DBT and DBE are compared with those based on the hierarchical clustering (HC) principle. For a second aim, this paper explores real-time applicability by testing the methods' ability to incorporate new time windows dynamically.
Keywords:
Traffic data
Origin-destination matrix
Spatiotemporal visualization
Correspondence analysis
Taxicab correspondence analysis
Principal component analysis
Hierarchical clustering

Journal

J
Journal of Transportation Engineering Part A-Systems
IF:
2.1
Papers:
123
Citations:
1.9K

Organization

No organization information available