Scaling GPS trajectories to match point traffic counts: A convex programming approach and Utah case study
| Autor: | Zachary Vander Laan, Nikola Marković, Seth Miller |
|---|---|
| Rok vydání: | 2020 |
| Předmět: |
050210 logistics & transportation
021103 operations research Scale (ratio) business.industry Computer science 05 social sciences 0211 other engineering and technologies Transportation 02 engineering and technology Overfitting 0502 economics and business Convex optimization Statistics Global Positioning System Point (geometry) Least absolute deviations State (computer science) Business and International Management business Scaling Civil and Structural Engineering |
| Zdroj: | Transportation Research Part E: Logistics and Transportation Review. 143:102105 |
| ISSN: | 1366-5545 |
| DOI: | 10.1016/j.tre.2020.102105 |
| Popis: | This paper considers the problem of inferring statewide traffic patterns by scaling massive GPS trajectory data, which capture about 3% of the overall traffic in Utah. It proposes a least absolute deviations model with controlled overfitting to scale 2.3 million trajectories such that resulting data best fit vehicle counts measured by 296 traffic sensors across the state. The proposed model improves on an often-cited approach from the literature and achieves 45% lower error for locations not seen in model training, obtaining 18% median hourly error across all test locations. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|
Full Text from ScienceDirect