Over the years, many TMCs have accumulated large loop detector datasets due to the loops' extensive presence in existing infrastructure. However, the information in these data sets often lies dormant partly because of the lack of effective means to summarize and display it. We present a simple visualization technique developed for loop detector data which addresses this issue. The method uses color-encoded images to visualize loop detector measurements as a function of space and time. We use examples based on a field dataset to illustrate some of its possible applications.

In this dissertation I explore and develop statistical methodology motivated by the scientific goals and engineering constraints of the Taiwan-America-Occultation-Survey (TAOS), a large astronomy survey dedicated to understanding populations in our outer solar system. I derive minimal conditions under which detection procedures are valid and robust, discuss the filtering and diagnostics necessary to meet those conditions, and explain how detection results lead to inference about population parameters. I also develop and compare several parametric and non-parametric detection methods for use with TAOS data, and also for the proposed TAOS II survey, which will have higher-resolution data and lower noise levels.

In this paper, we conduct an empirical comparison of travel time estimation methods based on single-loop detector data. The methods of concern are the regression method based on an intuitive stochastic model as proposed by Petty et al. in [7], and the conventional method of using an identity relating speed, flow and occupancy with the assumption of a common vehicle length. The analysis is tailored to fit in the limitations imposed by available field data sets. We also introduce several variations of the regression method and give examples which suggest directions for future work to further improve the regression method. The comparison is composed of three interrelated parts, each with a different focus:local comparison (concerning a single link of freeway), comparison of estimated section travel times over a prolonged stretch of freeway with multiple links and a visualized approach which enables investigation of performance patterns in time and space of the estimation methods.

We study prediction when features are observed with measurement error. The research is motivated by classification challenges in astronomy.

In Chapter 1 we introduce the periodic variable star classification problem. Periodic variable stars are periodic functions which belong to a particular physical class. These functions are often sparsely sampled, which introduces measurement error when attempting to estimate period, amplitude, and other function features. We discuss how measurement error can impact performance of periodic variable star classifiers. We introduce two general strategies, noisification and denoisification, for addressing measurement error in prediction problems.

In Chapter 2 we study density estimation with Berkson error. In this problem, one observes a sample from the density $f_X$ and seeks to estimate $f_Y$, the convolution of $f_X$ with a known error distribution. We derive asymptotic results for the behavior of the mean integrated squared error for kernel density estimates of $f_Y$. The presence of error generally increases convergence rates of estimators and optimal smoothing parameters. We briefly discuss some potential applications for this work, including classification tasks involving measurement error.

In Chapter 3 we study prediction of a continuous response for an observation with measurement error in its features. Using Nadaraya Watson type estimators we derive limit theorems for convergence of the mean squared error as a function of the smoothing parameters.

In Chapter 4 we study the effects of measurement error on classifier performance using data from the Optical Gravitational Lensing Experiment (OGLE) and the Hipparcos satellite. We illustrate some challenges in constructing statistical classifiers when the training data is collected by one astronomical survey and the unlabeled data is collected by a different survey. We use noisification to construct classifiers that are robust to some sources of measurement error and training--unlabeled data set differences.

We present a novel, off-line approach for evaluating incident detection algorithms. Previous evaluations have focused on determining the detection rate versus false alarm rate curve -- a process which we argue is inherently fraught with difficulties. Instead, we propose a cost-benefit analysis where cost mimics the real costs of implementing the algorithm and benefit is in terms of reduction in congestion. We argue that these quantities are of more practical interest than the traditional rates. Moreover, these costs, estimated on training data, can be used both as a mechanism to fine-tune a single algorithm as well as a meaningful quantity for direct comparisons between different types of incident detection algorithms. We demonstrate our approach with a detailed example. Key words: Incident detection