Statistical topics in transportation studies.

摘要

In recent years, the advances in traffic surveillance technologies have resulted in huge amounts of transportation data. The dissertation covers several topics concerned with understanding a particular type of traffic data—those collected by loop detectors, and with studying travel time estimation and prediction methods using such data.; We have been motivated to study the loop detector data because it has become the predominant source of traffic data, due to its low cost and technological maturity. The dissertation is consisted of three parts. We first present an informative visualization technique to mine large loop detector data sets in Chapter 3. Chapter 4 describes an empirical evaluation of travel time estimation methods using single-loop data. Lastly, we propose a short-term travel time prediction method using a linear model with time-varying coefficients in Chapter 5.; All results in the dissertation are based on real loop detector data sets. One of the contributions of the dissertation is to present examples of dealing with data quality issues in loop detector data.