Spatial Copula Model for Imputing Traffic Flow Data from Remote Microwave Sensors.

Area of Science:

• Transportation Engineering
• Data Science
• Spatial Statistics

Background:

• Missing data in traffic sensors is a growing problem, with up to 50% of values missing in some networks.
• Accurate imputation of missing traffic flow data is crucial but challenging due to data gaps.

Purpose of the Study:

• To propose and validate copula-based models for spatial interpolation of traffic flow data.
• To address limitations of existing methods that struggle with anomalous traffic patterns.

Main Methods:

• Developed copula-based models to connect correlation functions and marginal distribution functions of traffic flow.
• Compared the performance of copula-based models against three kriging methods using real-world traffic sensor data.

Main Results:

• Copula-based models demonstrated superior performance in imputing missing traffic flow data compared to kriging methods.
• The models were particularly effective on roads exhibiting irregular traffic flow patterns.

Conclusions:

• Copula-based models offer a robust and effective approach for spatial interpolation of traffic data.
• These models show significant potential for addressing missing data challenges in extensive transportation networks.