Master of Science / Department of Statistics / Perla Reyes / Heterodera glycines, or the soybean cyst nematode (SCN), is a parasite that targets and damages the roots of soybean plants. It is the most yield-limiting pathogen of soybean in the U.S. and the reliable detection and accurate estimation of population densities is crucial to research and management of this pathogen. A study was performed to understand the effects of crop rotation on the prevalence of SCN. Standard sampling procedures in the plant pathology community dictate taking soil samples from potentially infected fields, processing them and counting the number of eggs in one 1 mL subsample via microscope. Suspecting the traditional procedure may lead to invalid results, false negatives in particular, the researcher created and implemented a sampling procedure based on his knowledge of sampling methods and constraints of sampling in the field. Using the data collected, we will discuss the strengths and limitations of the procedure in estimating the population density of SCN in the field. In addition, a simulation study informed by the data will be conducted to determine a sampling strategy that will yield accurate results while still considering the conditions in the field. Knowledge on how the different stages of the sampling procedure for SCN affect the accurate detection of the pathogen would extend to experimental designs and sampling methodologies for other soil dwelling organisms.

The Low Income Housing Tax Credit Program is the primary federal program for producing new units of affordable housing. The program provides financial incentives to private developers to develop and operate affordable rental housing. In recent years, evidence has emerged that the program has led to clusters of subsidized housing in some cities. It is hardly surprising that some clustering would exist in a program in which the housing is constructed and owned by private developers. Despite the significant number of units produced by the program and despite the potential tendency for clustering of units built under this program, the locational patterns within the LIHTC program remain largely unexamined. Instead, most studies of the LIHTC program have focused on the national level rather than on individual cities. In contrast to previous studies, this study seeks to improve our understanding of variations in the LIHTC program across cities. The hypothesis of this study is that, because private developers produce housing in the LIHTC program and because the factors that influence private developers vary across cities, there is likely to be significant variation in the locational patterns of LIHTC developments across cities. The results of this study show, among other things, that clustering of LIHTC properties exists in the study cities, this clustering is extreme in some cases, and the clusters are associated with high poverty tracts in some cities. Given the LIHTC program's emphasis on market-driven policies and a similar emphasis in some other federal housing programs, such findings will likely be applicable to other affordable housing programs.

A citricultura é uma das principais atividades agrícolas do Brasil e o estado de São Paulo concentra a maior área produtora de laranjas do mundo. O conhecimento de padrões da incidência de doenças cítricas no tempo e no espaço é relevante para o setor e permite a descrição da dinâmica dessas doenças, podendo indicar estratégias para controle de epidemias. Neste trabalho são consideradas duas doenças que afetam a cultura de citros, a leprose e a morte súbita dos citros utilizando dados provenientes do monitoramento de talhões. Um aspecto relevante para estudos de doenças como a leprose dos citros, considerada uma grave virose na citricultura brasileira, é a investigação do padrão espacial e dos efeitos temporais da sua incidência dentro do talhão. Métodos exploratórios para determinar se o padrão espacial é ou não agregado são frequentemente utilizados. Entretanto é possível explorar e descrever os dados adotando um modelo explícito, permitindo discriminar e quantificar os efeitos através de parâmetros para co-variáveis que representam os aspectos de interesse. Uma das alternativas é a adoção de modelos autologísticos, que estendem o modelo de regressão logística para acomodar efeitos espaciais. Para implementar esse modelo é necessário que se reuse os dados para extrair co-variáveis espaciais, o que requer extensões na metodologia e algoritmos para avaliar adequadamente a variância das estimativas. Neste trabalho utiliza-se o modelo autologístico na análise de dados de incidência de doenças em plantas cítricas coletados em pontos referenciados no espaço e no tempo em um talhão. é mostrado como o modelo autologístico é apropriado para investigar doenças desse tipo, bem como é feita uma descrição do modelo e dos aspectos computacionais necessários para a estimação do modelo. São abordados métodos de seleção e avaliação de modelos autologísticos que relacionam fatores que afetam a disseminação da doença com padrões espaciais e efeitos temporais. Desta forma é possível realizar avaliações objetivas dos efeitos dos fatores considerados sobre a incidência da doença através dos parâmetros estimados do modelo proposto e quantificar o efeito da presença da doença em diferentes estruturas de vizinhança. A modelagem dos dados identificou dependência espacial entre as plantas e o modelo adotado permitiu quantificar as variações na probabilidade de doença em função do status das plantas na vizinhança. A metodologia apresentada aqui não se restringe a cultura de citros pode ser usada na avaliação de padrões espaço-temporais e efeitos de fatores que afetem doenças de plantas em condições semelhantes. / The citrus industry is a major agricultural activity in Brazil and the state of Sao Paulo concentrates the largest production area of oranges in the world. The knowledge of incidence patterns of citrus diseases in time and space is relevant to the industry and allows the description of the dynamics of these diseases and may indicate strategies for epidemics control. In this work are considered two diseases that affect the cultivation of citrus, leprosis and citrus sudden death using data from the tracking of stands. An important point when studying diseases such as the citrus leprosis, considered a serious viral disease in the Brazilian citrus industry, is the investigation of the spatial pattern and temporal effects of the disease incidence within a stand. Exploratory methods to determine if the spatial pattern is or not added are frequently used. However it is possible to explore and describe the data adopting an explicit model, allowing to discriminate and quantify the effects through parameters for covariates that represent aspects of interest. To implement this model is necessary to reuse the data in order to extract spatial covariates, which requires extensions in the methodology and algorithms to assess properly the variance of estimates. In this work, the autologistic model is used in the analysis of diseases incidence data in citrus plants collected in points referenced in space and time in a stand. It is shown how the autologistic model is appropriate to investigate such diseases, and there is a description of the model and computational aspects needed to estimate the model. Thus it is possible to achieve objective assessments of the effects of the factors considered on the incidence of the disease through the estimated parameters of the proposed model and quantify the disease presence effects in different neighborhood structures. The modeling of the data has identified a spatial dependence between the plants and the adopted model allowed to quantify the changes in the probability of disease according to the status of the plants in the neighbourhood. The methodology presented here is not restricted to the cultivation of citrus. It can be used in the assessment of spatial-temporal patterns and effects of factors that affect the diseases in plants under similar conditions.

O estudo da distribuição de pragas em espaço e tempo em sistemas agrícolas fornece informação importante sobre os mecanismos de dispersão das espécies e sua interação com fatores ambientais. Esse tipo de estudos também é de muita ajuda no desenvolvimento de planos de amostragem, na otimização de programas de manejo integrado de pragas e no planejamento de experimentos. O objetivo deste trabalho foi comparar vários modelos hierárquicos na modelagem da variação espaço-temporal da infestação da broca do café visando produzir mapas de risco da infestação que descrevam adequadamente o processo de infestação. Foram usadas diferentes combinações de efeitos aleatórios representando variabilidade não estruturada, com diferentes escolhas de distribuições a priori para os parâmetros e os hiperparâmetros dos modelos. Foram também usados diferentes esquemas de vizinhança para representar a correlação espacial dos dados. O ajuste dos modelos foi feito usando métodos MCMC. A estatística deviance e funções de perda quadrática foram usadas para a comparação entre modelos. Os resultados são apresentados como uma seqüência de mapas de risco de infestação. / Study of agricultural pests distribution in space and time provides important information about the species dispersion mechanisms and its interaction with environmental factors. It also helps the development of sampling plans, the integrated pest management and planning of experiments. The aim of this work was to compare several hierarchical models in modelling the spatio-temporal variation of the coffee berry borer infestation in order to produce risk maps. Different combinations of random effects representing spatially structured and unstructured variability were used, with different prior distributions for the parameters and hyperparameters. Also different neighbourhood schemes were used to represent the spatial correlation of the data. The model fitting was done using MCMC methods and deviance and squared loss function were used for the comparison between models. The results are presented as a sequence of risk maps.

A citricultura é uma das principais atividades agrícolas do Brasil e o estado de São Paulo concentra a maior área produtora de laranjas do mundo. O conhecimento de padrões da incidência de doenças cítricas no tempo e no espaço é relevante para o setor e permite a descrição da dinâmica dessas doenças, podendo indicar estratégias para controle de epidemias. Neste trabalho são consideradas duas doenças que afetam a cultura de citros, a leprose e a morte súbita dos citros utilizando dados provenientes do monitoramento de talhões. Um aspecto relevante para estudos de doenças como a leprose dos citros, considerada uma grave virose na citricultura brasileira, é a investigação do padrão espacial e dos efeitos temporais da sua incidência dentro do talhão. Métodos exploratórios para determinar se o padrão espacial é ou não agregado são frequentemente utilizados. Entretanto é possível explorar e descrever os dados adotando um modelo explícito, permitindo discriminar e quantificar os efeitos através de parâmetros para co-variáveis que representam os aspectos de interesse. Uma das alternativas é a adoção de modelos autologísticos, que estendem o modelo de regressão logística para acomodar efeitos espaciais. Para implementar esse modelo é necessário que se reuse os dados para extrair co-variáveis espaciais, o que requer extensões na metodologia e algoritmos para avaliar adequadamente a variância das estimativas. Neste trabalho utiliza-se o modelo autologístico na análise de dados de incidência de doenças em plantas cítricas coletados em pontos referenciados no espaço e no tempo em um talhão. é mostrado como o modelo autologístico é apropriado para investigar doenças desse tipo, bem como é feita uma descrição do modelo e dos aspectos computacionais necessários para a estimação do modelo. São abordados métodos de seleção e avaliação de modelos autologísticos que relacionam fatores que afetam a disseminação da doença com padrões espaciais e efeitos temporais. Desta forma é possível realizar avaliações objetivas dos efeitos dos fatores considerados sobre a incidência da doença através dos parâmetros estimados do modelo proposto e quantificar o efeito da presença da doença em diferentes estruturas de vizinhança. A modelagem dos dados identificou dependência espacial entre as plantas e o modelo adotado permitiu quantificar as variações na probabilidade de doença em função do status das plantas na vizinhança. A metodologia apresentada aqui não se restringe a cultura de citros pode ser usada na avaliação de padrões espaço-temporais e efeitos de fatores que afetem doenças de plantas em condições semelhantes. / The citrus industry is a major agricultural activity in Brazil and the state of Sao Paulo concentrates the largest production area of oranges in the world. The knowledge of incidence patterns of citrus diseases in time and space is relevant to the industry and allows the description of the dynamics of these diseases and may indicate strategies for epidemics control. In this work are considered two diseases that affect the cultivation of citrus, leprosis and citrus sudden death using data from the tracking of stands. An important point when studying diseases such as the citrus leprosis, considered a serious viral disease in the Brazilian citrus industry, is the investigation of the spatial pattern and temporal effects of the disease incidence within a stand. Exploratory methods to determine if the spatial pattern is or not added are frequently used. However it is possible to explore and describe the data adopting an explicit model, allowing to discriminate and quantify the effects through parameters for covariates that represent aspects of interest. To implement this model is necessary to reuse the data in order to extract spatial covariates, which requires extensions in the methodology and algorithms to assess properly the variance of estimates. In this work, the autologistic model is used in the analysis of diseases incidence data in citrus plants collected in points referenced in space and time in a stand. It is shown how the autologistic model is appropriate to investigate such diseases, and there is a description of the model and computational aspects needed to estimate the model. Thus it is possible to achieve objective assessments of the effects of the factors considered on the incidence of the disease through the estimated parameters of the proposed model and quantify the disease presence effects in different neighborhood structures. The modeling of the data has identified a spatial dependence between the plants and the adopted model allowed to quantify the changes in the probability of disease according to the status of the plants in the neighbourhood. The methodology presented here is not restricted to the cultivation of citrus. It can be used in the assessment of spatial-temporal patterns and effects of factors that affect the diseases in plants under similar conditions.

This dissertation seeks to extend quantitative research on environmental justice and address methodological limitations of previous studies by: (a) using new indicators of exposure to air pollution and contemporary risk modeling techniques; (b) assessing disparities in human health risks, instead of focusing only on potential exposure or proximity to pollution sources; and (c) using multivariate regression models that consider the effects of spatial dependence. The case study examines racial/ethnic and socioeconomic disparities in the geographic distribution of exposure to airborne toxic emissions from industrial point sources in the Houston-Galveston-Brazoria metropolitan statistical area. Industrial pollution sources for this study comprise facilities listed in the US EPA's Toxic Release Inventory (TRI). The Risk-Screening Environmental Indicator (RSEI) model is used to estimate potential human health risks from air pollutants based on data on toxicity and dispersion of chemical releases from TRI facilities. The analyses utilize four indicators of potential exposure to industrial pollution: (a) presence or absence of air emissions, (b) total quantities (pounds) of air emissions, (c) toxicity-weighted quantities of emissions and (d) modeled risk scores based on the cumulative health risk posed by air emissions. Traditional linear regression and spatial autoregressive techniques based on several neighborhood configurations are used to model the occurrence and magnitude of these four indicators, using relevant explanatory variables from the 2000 census, at the census tract and block groups levels of aggregation. Results indicate a disproportionate pattern of health risks from TRI facilities in the HGB-MSA, with the Hispanic population facing the highest exposure. The locations and magnitude of toxic pollution are significantly statistical effected by the presence of minority residents and population density. Additionally, key differences in the significance of explanatory variables between the spatial and conventional regression models demonstrate the importance of correcting for spatial dependence in environmental justice analysis. The analytical results for several variables are also sensitive to the choice of data resolution (tract or block group). Overall, this study indicates that more flexible spatial analytic techniques are required to improve the identification of environmental injustice and past studies utilizing conventional statistical methods should be revisited to explicitly account for spatial effects.

Response-Adaptive (RA) designs are used to adaptively allocate patients in clinical trials. These methods have been generalized to include Covariate-Adjusted Response-Adaptive (CARA) designs, which adjust treatment assignments for a set of covariates while maintaining features of the RA designs. Challenges may arise in multi-center trials if differential treatment responses and/or effects among sites exist. We propose Site-Adjusted Response-Adaptive (SARA) approaches to account for inter-center variability in treatment response and/or effectiveness, including either a fixed site effect or both random site and treatment-by-site interaction effects to calculate conditional probabilities. These success probabilities are used to update assignment probabilities for allocating patients between treatment groups as subjects accrue. Both frequentist and Bayesian models are considered. Treatment differences could also be attributed to differences in social determinants of health (SDH) that often manifest, especially if unmeasured, as spatial heterogeneity amongst the patient population. In these cases, patient residential location can be used as a proxy for these difficult to measure SDH. We propose the Location-Adjusted Response-Adaptive (LARA) approach to account for location-based variability in both treatment response and/or effectiveness. A Bayesian low-rank kriging model will interpolate spatially-varying joint treatment random effects to calculate the conditional probabilities of success, utilizing patient outcomes, treatment assignments and residential information. We compare the proposed methods with several existing allocation strategies that ignore site for a variety of scenarios where treatment success probabilities vary.

In this dissertation, we explore the properties of correlation structure for spatio-temporal point processes and a quantitative spatial process. Spatio-temporal point processes are often assumed to be separable; we propose a formal approach for testing whether a particular data set is indeed separable. Because of the resampling methodology, the approach requires minimal conditions on the underlying spatio-temporal process to perform the hypothesis test, and thus is appropriate for a wide class of models. Africanized Honey Bees (AHBs, Apis mellifera scutellata) abscond more frequently and defend more quickly than colonies of European origin. That they also utilize smaller cavities for building colonies expands their range of suitable hive locations to common objects in urban environments. The aim of the AHB study is to create a model of this quantitative spatial process to predict where AHBs were more likely to build a colony, and to explore what variables might be related to the occurrences of colonies. We constructed two generalized linear models to predict the habitation of water meter boxes, based on surrounding landscape classifications, whether there were colonies in surrounding areas, and other variables. The presence of colonies in the area was a strong predictor of whether AHBs occupied a water meter box, suggesting that AHBs tend to form aggregations, and that the removal of a colony from a water meter box may make other nearby boxes less attractive to the bees.

Primary health care (PHC) is a term used to refer to the parts of the health system that people interact with most of the time when health care is needed. It is considered the first point of contact for health services in Canada. Access to PHC services is an important issue regarding health care delivery in Canada today. There is a need to advance current understanding of access to PHC providers at local scales such as neighbourhoods. The primary objective of this study is to examine the variation in geographic (spatial) accessibility to permanently located primary care services in the Canadian urban environment. Furthermore, the analysis of spatial patterns of accessibility, both visually and statistically using GIS, is to provide a better understanding of among and between neighbourhood variations. This research took place in the 14 urban areas across Canada: Victoria and Vancouver, British Columbia; Calgary and Edmonton, Alberta; Saskatoon, Saskatchewan; Winnipeg, Manitoba; Hamilton, and Toronto, Ontario; Montréal and Québec, Quebec; Halifax, Nova Scotia; St. John’s, Newfoundland; Saint John, New Brunswick; and Ottawa–Gatineau, Ontario and Quebec. A GIS based method, the Three-Step Floating Catchment Area (3SFCA), was applied to determine the spatial accessibility to PHC services (accessibility score). First, for increasing geocoding match rates with reduced positional uncertainty, an integrated geocoding technique was developed after an empirical comparison of the geocoding results based on manually built and online geocoding services and subsequently applied to generate geographic coordinates of PHC practices which are an essential element for measuring potential access to health care. Next, the results of the Three-Step Floating Catchment Area (3SFCA) method was compared with simpler approachs to calculate the City level physician-to-population ratios and this research highlights the benefit of using the 3SFCA method over simpler approaches in urban areas by providing similar or comparable results of City level physician-to-population ratios with the advantage of intra-urban measurements. Further, the results point out that considerable spatial variation in geographical accessibility to PHC services exists within and across Canadian urban areas and indicate the existence of clusters of poorly served neighbourhoods in all urban areas. In order to investigate the low accessibility scores in relation to population health care needs, spatial statistical modeling techniques were applied that revealed variations in geographical accessibility to PHC services by comparing the accessibility scores to different socio-demographic characteristics across Canadian urban settings. In order to analyse how these relationships between accessibility and predictors vary at a local scale within an urban area, a local spatial regression technique (i.e., geographically weighted regression or GWR) was applied in two urban areas. The results of GWR modelling demonstrates intra-urban variations in the relationships between socio-demographic variables and the geographic accessibility to PHC services. In addition, the influences of “unit of analysis” on accessibility score were analyzed using spatial statistical modeling that emphasize the use of units of analysis that are pertinent to policy and planning purposes such as city defined neighbourhoods. Overall, this research shows the importance of measuring geographic accessibility of PHC services at local levels for decision makers, planners, researchers, and policy makers in the field of public health and health geography. This dissertation will advance current understanding of access to primary care in Canadian urban settings from the perspective of the neighbourhood.

Many real-world datasets have spatial components that provide valuable information about characteristics of the data. In this dissertation, a novel framework for adaptive models that exploit spatial information in data is proposed. The proposed framework is mainly based on development and applications of Gaussian processes. First, a supervised learning method is proposed for the classification of hyperspectral data with spatially adaptive model parameters. The proposed algorithm models spatially varying means of each spectral band of a given class using a Gaussian process regression model. For a given location, the predictive distribution of a given class is modeled by a multivariate Gaussian distribution with spatially adjusted parameters obtained from the proposed algorithm. The Gaussian process model is generally regarded as a good tool for interpolation, but not for extrapolation. Moreover, the uncertainty of the predictive distribution increases as the distance from the training instances increases. To overcome this problem, a semi-supervised learning algorithm is presented for the classification of hyperspectral data with spatially adaptive model parameters. This algorithm fits the test data with a spatially adaptive mixture-of-Gaussians model, where the spatially varying parameters of each component are obtained by Gaussian process regressions with soft memberships using the mixture-of-Gaussian-processes model. The proposed semi-supervised algorithm assumes a transductive setting, where the unlabeled data is considered to be similar to the training data. This is not true in general, however, since one may not know how many classes may existin the unexplored regions. A spatially adaptive nonparametric Bayesian framework is therefore proposed by applying spatially adaptive mechanisms to the mixture model with infinitely many components. In this method, each component in the mixture has spatially adapted parameters estimated by Gaussian process regressions, and spatial correlations between indicator variables are also considered. In addition to land cover and land use classification applications based on hyperspectral imagery, the Gaussian process-based spatio-temporal model is also applied to predict ground-based aerosol optical depth measurements from satellite multispectral images, and to select the most informative ground-based sites by active learning. In this application, heterogeneous features with spatial and temporal information are incorporated together by employing a set of covariance functions, and it is shown that the spatio-temporal information exploited in this manner substantially improves the regression model. The conventional meaning of spatial information usually refers to actual spatio-temporal locations in the physical world. In the final chapter of this dissertation, the meaning of spatial information is generalized to the parametrized low-dimensional representation of data in feature space, and a corresponding spatial modeling technique is exploited to develop a nearest-manifold classification algorithm. / text