Spatial-Temporal Congestion Identification Based on Time Series Similarity Considering Missing Data

Area of Science:

• Transportation Science
• Network Analysis
• Data Science

Background:

• Traffic congestion is a dynamic phenomenon varying in space and time.
• Existing methods for traffic flow analysis often struggle with missing data, leading to inaccuracies.
• Understanding congestion dynamics is crucial for network performance and bottleneck identification.

Purpose of the Study:

• To develop a new traffic flow operational index time series similarity measurement.
• To establish a method for identifying dynamic network bottlenecks, even with missing data.
• To improve the accuracy of traffic congestion analysis.

Main Methods:

• A two-stage similarity measurement process (first and second order) is proposed to handle missing data.
• Traffic flow operational similarity is calculated between neighboring road links.
• Similarity results are used to cluster spatial-temporal congestion patterns.

Main Results:

• The developed method effectively handles missing traffic flow data.
• The approach successfully identifies dynamic network bottlenecks.
• Results align with empirical observations and offer new insights into traffic flow.

Conclusions:

• The proposed similarity measurement is a robust approach for analyzing traffic flow data with missing values.
• This method provides a reliable basis for identifying dynamic network bottlenecks.
• The findings contribute to a better understanding of traffic congestion dynamics.