Most estimates of wood supply do not reflect the true availability of wood resources. The availability of wood resources ultimately depends on collective wood harvesting decisions across the landscape. Both social and biophysical constraints impact harvesting decisions and thus the availability of wood resources. While most constraints do not completely inhibit harvesting, they may significantly reduce the probability of harvest. Realistic assessments of woody availability and distribution are needed for effective forest management and planning. This study focuses on predicting the probability of harvest at forested FIA plot locations in Virginia. Classification and regression trees, conditional inferences trees, random forest, balanced random forest, conditional random forest, and logistic regression models were built to predict harvest as a function of social and biophysical availability constraints. All of the models were evaluated and compared to identify important variables constraining harvest, predict future harvests, and estimate the available wood supply. Variables related to population and resource quality seem to be the best predictors of future harvest. The balanced random forest and logistic regressions models are recommended for predicting future harvests. The balanced random forest model is the best predictor, while the logistic regression model can be most easily shared and replicated. Both models were applied to predict harvest at recently measured FIA plots. Based on the probability of harvest, we estimate that between 2012 and 2017, 10 – 21 percent of total wood volume on timberland will be available for harvesting. / Master of Science

Motivated by real-time monitoring of dialysis, we aim at detecting difference between groups of Raman spectra generated from dialyzates at different time in one session. Baseline correction being a critical procedure in use of Raman Spectra, existing methods may not perform well on dialysis spectra due to nature of dialyzates, which contain numerous chemicals compounds. We first developed a new baseline correction method, Iterative Smoothing-spline with Root Error Adjustment (ISREA), which automatically adjusts intensities and employs smoothing-spline to produce a baseline in each iteration, providing better performance on dialysis spectra than a popular method Goldindec, and better accuracy regardless of types of samples. We proposed a two sample hypothesis testing on groups of baseline-corrected Raman spectra with ISREA. The uniqueness of the test lies in nature of the tested data. Instead of using Raman spectra as curves, we also consider a vector whose elements are peak intensities of biomarkers, meaning the data is regarded as mixed data and that a spectrum curve and a vector compose one observation. Our method tests on equality of the means of the two groups of mixed data. This method is based on asymptotic properties of the covariance of mixed data and FPCA. Simulation studies shows that our method is applicable to small sample size with proper power and size control. Meanwhile, to locate regions that contribute most to significant difference between two groups of univariate functional data, we developed a method to estimate the a sparse coefficient function by using a L1 norm penalty in functional logistic regression, and compared its performance with other methods. / Doctor of Philosophy / In U.S., there are more than 709,501 patients with End-Stage Renal Disease (ESRD). For those patients, dialysis is a standard treatment. While dialysis is time-consuming, expensive, and uncomfortable, it requires patients to take three sessions every week in facilities, and each session lasts for four hours regardless of patients' condition. An affordable, fast, and widely-applied technique called Raman spectroscopy draws attention. Spectral data from used dialysate samples collected at different time in one session can give information on the dialysis process and thus make real-time monitoring possible. With spectral data, we want to develop a statistical method that helps real-time monitoring on dialysis. This method can provide physicians with statistical evidence on dialysis process to improve their decision making, therefore increases efficiency of dialysis and better serve patients. On the other hand, Raman spectroscopy demands preprocessing called baseline correction on the raw spectra. A baseline is generated because of the nature of Raman technique and its instrumentation, which adds complexity to the spectra and interfere with analysis. Despite popularity of this technique and many existing baseline correction method, we found performance on dialysate spectra under expectation. Hence, we proposed a baseline correction method called Iterative Smoothing-spline with Root Error Adjustment (ISREA) and ISREA can provide better performance than existing methods. In addition, we come up with a method that is able to detect difference between the two groups of ISREA baseline-corrected spectra from dialysate collected at different time. Furthermore, we proposed and applied sparse functional logistic regression on two groups to locate regions where the significant difference comes from.

Headwater streams change in wet length in response to storm events and seasonal moisture conditions. These low-order channels with temporary flow are pervasive across arid and humid environments yet receive little attention in comparison to perennial waterways. This dissertation examines headwater stream length dynamics at multiple spatial and temporal scales across the Appalachians. I mapped wet stream length in four Appalachian physiographic provinces--the Appalachian Plateau, Blue Ridge, New England, and Valley and Ridge--to characterize seasonal expansion and contraction of the wet network at a broad, regional scale. Conversely, most existing field studies of stream length in headwaters are limited to a single study area or geographic setting. Field mappings showed that wet stream length varies widely within the Appalachians; network dynamics correlated with regional geology as well as local site lithology, geologic structure, and the depth, size, and spatial distribution of surficial sediment deposits. I used the field data to create logistic regression models of the wet network in each physiographic province at high and low runoffs. Topographic metrics derived from elevation data were able to explain the discontinuous pattern of headwater streams at different flow conditions with high classification accuracy. Finally, I used flow intermittency sensors in a single Valley and Ridge catchment to record channel wetting and drying at a high temporal resolution. The sensors indicated stream length hysteresis during storms with low antecedent moisture, with a higher wet network proportion on the rising limb than on the falling limb of events. As a result, maximum network extension can precede peak runoff by minutes to hours. Accurate maps of headwater streams and an understanding of wet network dynamics through time are invaluable for applications surrounding watershed management and environmental policy. These findings will contribute to the burgeoning research on temporary streams and are additionally relevant for studies of runoff generation, biogeochemical cycling, and mass fluxes of material from headwaters. / Ph. D.

Fixed-object crashes have long time been considered as major roadway safety concerns. While previous relevant studies tended to address such crashes in the context of roadway departures, and heavily relied on police-reported accidents data, this study integrated the SHRP2 NDS and RID data for analyses, which fully depicted the prior to, during, and after crash scenarios. A total of 1,639 crash, near-crash events, and 1,050 baseline events were acquired. Three analysis methods: logistic regression, support vector machine (SVM) and artificial neural network (ANN) were employed for two responses: crash occurrence and severity level. Logistic regression analyses identified 16 and 10 significant variables with significance levels of 0.1, relevant to driver, roadway, environment, etc. for two responses respectively. The logistic regression analyses led to a series of findings regarding the effects of explanatory variables on fixed-object event occurrence and associated severity level. SVM classifiers and ANN models were also constructed to predict these two responses. Sensitivity analyses were performed for SVM classifiers to infer the contributing effects of input variables. All three methods obtained satisfactory prediction performance, that was around 88% for fixed-object event occurrence and 75% for event severity level, which indicated the effectiveness of NDS event data on depicting crash scenarios and roadway safety analyses. / Master of Science / Fixed-object crashes happen when a single vehicle strikes a roadway feature such as a curb or a median, or runs off the road and hits a roadside feature such as a tree or utility pole. They have long time been considered as major highway safety concerns due to their high frequency, fatality rate, and associated property cost. Previous studies relevant to fixed-object crashes tended to address such crashes in the contexture of roadway departures, and heavily relied on police-reported accident data. However, many fixed-object crashes involved objects in roadway such as traffic control devices, roadway debris, etc. The police-reported accident data were found to be weak in depicting scenarios prior to, during crashes. Also, many minor crashes were often kept unreported. The Second Strategic Highway Research Program (SHRP2) Naturalistic Driving Study (NDS) is the largest NDS project launched across the country till now, aimed to study driver behavior or, performance-related safety problems under real-world scenarios. The data acquisition systems (DASs) equipped on participated vehicles collect vehicle kinematics, roadway, traffic, environment, and driver behavior data continuously, which enable researchers to address such crash scenarios closely. This study integrated SHRP2 NDS and roadway information database (RID) data to conduct a comprehensive analysis of fixed-object crashes. A total of 1,639 crash, near-crash events relevant to fixed objects and animals, and 1,050 baseline events were used. Three analysis methods: logistic regression, support vector machine (SVM) and artificial neural network (ANN) were employed for two responses: crash occurrence and severity level. The logistic regression analyses identified 16 and 10 variables with significance levels of 0.1 for fixed-object event occurrence and severity level models respectively. The influence of explanatory variables was discussed in detail. SVM classifiers and ANN models were also constructed to predict the fixed-object crash occurrence and severity level. Sensitivity analyses were performed for SVM classifiers to infer the contributing effects of input variables. All three methods achieved satisfactory prediction accuracies of around 88% for crash occurrence prediction and 75% for crash severity level prediction, which suggested the effectiveness of NDS event data on depicting crash scenarios and roadway safety analyses.

The spatial distributions of four plant species native to Asia, yet considered invasive in southwestern Virginia, were studied in order to produce predictive habitat maps. The study took place in the mountains to the north of Blacksburg, VA, on National Forest lands. A random GPS survey of each of the four species, Microstegium vimineum, Lonicera japonica, Rosa multiflora and Elaeagnus umbellate, was used in combination with a series of Geographic Information System (GIS) layers representing environmental variables (Elevation, Aspect, Roads, Trails, Streams, & Normalized Difference Moisture Index) to produce logistic regression models. After field- validating the models, the models were ranked according to usefulness, with the E. umbellate model proving most accurate. It is hoped that such GIS models will allow forest managers to more productively search for invasive species within their jurisdiction, by indicating sites more likely to provide habitat suitable to the invasive species described by the model. A non-GIS search for correlations between the study species' presence and field-collected discrete environmental variables was also included. Both Disturbance and Canopy Cover were considered for their effect upon Microstegium vimineum, Lonicera japonica, Rosa multiflora and Elaeagnus umbellate presence. Using Pearson's Correlation with the Canopy Cover data, and Chi-squared Correlation with the Disturbance data, only R. multiflora and E. umbellate showed significant correlation to decreasing canopy cover. / Master of Science

This study extends the research of Rodgers (1999) on the relationship between parenting processes and adolescent sexual risk-taking. Parenting behaviors considering were parental monitoring, parent-adolescent communication, and parenting styles. Sexual risk-taking was determined by assessing number of lifetime sexual partners as well as use of condoms during last sexual intercourse. A sample (n=286) of 9th-12th grade males and females who reporting having had sexual intercourse were separated into two groups-those engaging in low sexual risk-taking or high sexual risk-taking. Logistic regression analysis revealed gender differences in the relationship between parents' behaviors and adolescent sexual risk-taking. For females, parenting monitoring of the adolescent's after-school whereabouts was related to a decrease in the odds that a daughter would take sexual risks. For males, parental monitoring of whom the adolescent male goes out with was related to a decrease in the odds of a son taking sexual risks. Several significant interaction effects were also found. / Master of Science

Improved definition of risk factors for equine colic is necessary to develop effective interventions to reduce colic incidence. A one-year prospective study was conducted to estimate colic incidence and to identify risk factors. Farms with greater than 20 horses were randomly selected from two adjacent counties of Virginia and Maryland. Management information was recorded by questionnaire for 31 farms with 1427 horses. Owners kept calendars to record occurrence of specified events. Colic was reported by the owner when a horse exhibited signs of abdominal pain. The incidence of colic was 10.6 colic cases per 100 horse-years, based on 104 cases per 983.5 horse-years. Twenty-five deaths occurred from all causes, the proportional mortality rate of colic was 7/25 (28%). Risk factors were analyzed by logistic regression at the farm-level and the horse-level with farm as a random effects variable. No farm-level variables were significant. Significant horse variables were: age 2-10 years, odds ratio (OR)=2.8 (95% confidence interval, 1.2-6.5); previous colic, OR=3.6(1.9-6.8); changes in concentrate feeding during the year, OR=3.6(1.6-5.4); more than one change in hay feeding during the year, OR=2.1(1.2-3.8); feeding high levels of concentrate (>2.5 kg/day dry matter, OR=4.8(1.4-16), >5 kg/day dry matter, OR=6.3(1.8-22)); and vaccination with monocytic ehrlichiosis vaccine during the study, OR=2.0(1.8-22). Feeding whole grain with or without other concentrates had less risk than diets without whole grain included. Variables related to concentrate feeding frequency or concentrate type could be substituted for the concentrate level variable. A nested analysis examined risk for the time period following an event. The odds ratio was determined for the proportion of cases with an event within 14 days prior to the colic-date, relative to the proportion of horses without colic with an event within 14 days of a date chosen at random from the observation time. Weather events were analyzed for the three days before the colic or assigned date. Foaling was analyzed for three time periods: before, 0-60 and 60-150 days post-foaling. Significant events were recent vaccination, OR=3.31(1.9-6.0); recent transport, OR=3.3(1.2-5.5); 60-150 days post-foaling, OR=5.9(1.8-13); and recent fever, OR=20(2.5-169). Snow on the day of the colic, OR=2.8(1.0-7) and humidity <50% the day before the colic OR=1.6(1.0-2.9) were marginally significant. / Ph. D.

Doctor of Philosophy / Department of Physics / Dean Zollman / As part of a collaborative study of the science preparation of elementary school teachers, we investigated the quality of students’ reasoning and explored the relationship between sophistication of reasoning and the degree to which the courses were considered inquiry oriented. To probe students’ reasoning, we developed open-ended written content questions with the distinguishing feature of applying recently learned concepts in a new context. We devised a protocol for developing written content questions that provided a common structure for probing and classifying students’ sophistication level of reasoning. In designing our protocol, we considered several distinct criteria, and classified students’ responses based on their performance for each criterion. First, we classified concepts into three types: Descriptive, Hypothetical, and Theoretical and categorized the abstraction levels of the responses in terms of the types of concepts and the inter-relationship between the concepts. Second, we devised a rubric based on Bloom’s revised taxonomy with seven traits (both knowledge types and cognitive processes) and a defined set of criteria to evaluate each trait. Along with analyzing students’ reasoning, we visited universities and observed the courses in which the students were enrolled. We used the Reformed Teaching Observation Protocol (RTOP) to rank the courses with respect to characteristics that are valued for the inquiry courses. We conducted logistic regression for a sample of 18 courses with about 900 students and reported the results for performing logistic regression to estimate the relationship between traits of reasoning and RTOP score. In addition, we analyzed conceptual structure of students’ responses, based on conceptual classification schemes, and clustered students’ responses into six categories. We derived regression model, to estimate the relationship between the sophistication of the categories of conceptual structure and RTOP scores. However, the outcome variable with six categories required a more complicated regression model, known as multinomial logistic regression, generalized from binary logistic regression. With the large amount of collected data, we found that the likelihood of the higher cognitive processes were in favor of classes with higher measures on inquiry. However, the usage of more abstract concepts with higher order conceptual structures was less prevalent in higher RTOP courses.

Cultural competence is widely recognized as an essential strategy for reducing health disparities. As the United States' population becomes increasingly ethno-culturally diverse, these disparities are becoming even more pronounced. One particular challenge in this regard concerns overweight/obesity prevalence among American adults, as a disproportionately high number of racial and ethnic minority adults are classified as overweight or obese. Dietary behavior assessments are often utilized by health and human services professionals to obtain the data necessary to promote goals such as the reduction and elimination of overweight/obesity across all ethno-cultural groups. The primary objective of this research study was to develop, test, and evaluate a culturally-competent dietary behavior assessment instrument by effectively synthesizing qualitative methods from Cognitive Anthropology with appropriate survey research and quantitative statistical methods. Specifically, a quantitative methods triangle of hierarchical cluster analysis, binary logistic regression, and Poisson regression in conjunction with the free listing qualitative research technique from Cognitive Anthropology was explored as a possible combined methodological approach for researchers and public health professionals wishing to develop a comprehensive understanding of dietary behaviors at the local community level. Binary logistic regression and Poisson regression enabled the relationship between selected food categories and certain demographic/cultural indicators to be modeled, while hierarchical cluster analyses enabled modeling of the distinct patterns of food category groupings that comprise individuals' regular diet. Additionally, initial qualitative analyses of the raw data promoted an understanding of the influence that the local fast food and dine-in restaurant environment has on the dietary behaviors of the target population. The results of this study suggest that a quantitative methods triangle of hierarchical cluster analysis, binary logistic regression analysis, and Poisson regression analysis founded upon qualitative research principles has potential for use as a combined methodological approach for researchers and public health professionals wishing to develop a comprehensive understanding of dietary behaviors at the local community level. By employing these techniques, researchers can analyze individual dietary behaviors and eating patterns from a multifaceted perspective. In turn, public health professionals can develop community-based, cross-culturally relevant programs and interventions that are equally effective across all ethno-cultural groups in their target population.

Higher-level education is very important these days, which can create very high thresholds for admission on popular programs on certain universities. In order to know what grade will be needed to be admitted to a program, a student can look at the threshold from previous years. We explored whether it was possible to generate accurate predictions of what the future threshold would be. We did this by using well-established machine learning classification models and admission data from 14 years back covering all applicants to the Computer Science and Engineering Program at KTH Royal Institute of Technology. What we found through this work is that the models are good at correctly classifying data from the past, but not in a meaningful way able to predict future thresholds. The models could not make accurate future predictions solely based on grades of past applicants. / Eftergymnasiala studier är väldigt viktiga numera, vilket kan leda till mycket höga antagningskrav på populära program på vissa universitet och högskolor. För att veta vilket betyg som krävs för att komma in på en utbildning så kan studenten titta på gränsen från tidigare år och utifrån det gissa sig till vad gränsen kommer vara kommande år. Vi undersöker om det är möjligt att, med hjälp av väletablerade, klassificerande Maskininlärnings-modeller kunna förutse antagningsgränsen i framtiden. Vi tränar modellerna på data med antagningsstatistik som sträcker sig tillbaka 14 år med alla ansökningar till civilingenjörs-programmet Datateknik på Kungliga Tekniska Högskolan. Det vi finner genom detta arbete är att modellerna är bra på att korrekt klassificera data från tidigare år, men att de inte, på ett meningsfullt sätt, kan förutse betygsgränsen kommande år. Modellerna kan inte göra detta endast genom data på betyg från tidigare år.