Laura Lindley Tupper
Published: 2016
Total Pages: 252
Get eBook
This work explores several approaches to the classification of complex data types, including high-dimensional, spatial, and time series data. It contains three main sections, each demonstrating a different application area. In Chapter 2, we explore the behavior of wind speed over time, using the Eastern Wind Dataset published by the National Renewable Energy Laboratory. To assess differences in intra-day time series, we propose a functional distance measure, the band distance, which extends the band depth of Lopez-Pintado and Romo (2009). This measure emphasizes the shape of time series or functional observations relative to other members of a dataset, and allows clustering of observations without reliance on pointwise Euclidean distance. We also examine adjustments for seasonality and representations in the time-frequency domain. We demonstrate the utility of the new method in simulation studies and an application to the MOST power grid algorithm. Chapter 3 presents multivariate time series analysis motivated by applications in biophysics. Electric Cell-substrate Impedance Sensing measures the electrical behavior of a cell culture exposed to an alternating current; we can obtain linked time series by observing the changes in impedance over time at various AC frequencies. We explore a space of features derived from these time courses, toward classification of different types of cells based on their electrical behavior. In Chapter 4 we perform an exploratory analysis of a dataset provided by the Regional Transit Service of Rochester, NY, describing ridership on public buses. We describe the overall behavior of bus ridership, and proceed to a stop-level analysis. By examining daily patterns of ridership volume, we can find clusters of similar stops, and compare this form of similarity to other measures of stop similarity based on location and route information.