Spatial dependency is commonly measured and tested with Moran's I statistic. The question to be answered is, whether far-off objects affect this statistic and influence the test. Far-off objects are observations that are far apart from all other objects in the dataset, i.e. they do not have spatial links to other design points. In the paper different possibilities of treating such objects are discussed, and their influence on Moran's I and the corresponding spatial autocorrelation test is analysed. (author's abstract) / Series: Research Report Series / Department of Statistics and Mathematics

The red imported fire ant (Solenopsis invicta)(RIFA) is a major pest in the United States, causing serious economic and human costs. This study explored the feasibility of using digital aerial remote sensing in multispectral/multitemporal detection and mapping of RIFA mounds. Comparison of photointerpretive mound counts versus ground control counts was performed within two grass types, common Bermuda and tall fescue. Flights collecting digital image data occurred at three intervals in 2009, with ground truth data collected collaterally. Poisson regression count modeling was first utilized for analysis of both datasets. Moran's Index geospatial analysis was applied following the Poisson model. Outcomes in this study from these models demonstrate their ability to robustly support studies for tracking and control of RIFA or other pest populations. Additionally, in one location, type of grass cover appeared to affect detectability of mounds between the two methods.

In this paper I will give a brief and general overview of the characteristics of spatial data, why it is useful to use such data and how to use the information included in spatial data. The first question to be answered is: how to detect spatial dependency and spatial autocorrelation in data? Such effects can for instance be found by calculating Moran's I, which is a measure for spatial autocorrelation. The Moran's I is also the basis for a test for spatial autocorrelation (Moran's test). Once we found some spatial structure we can use special models and estimation techniques. There are two famous spatial processes, the SAR- (spatial autoregressive) and the SMA- (spatial moving average process) process, which are used to model spatial effects. For estimation of spatial regression models there are mainly two different possibilities, the first one is called spatial filtering, where the spatial effect is filtered out and standard techniques are used, the second one is spatial two stage least square estimation. Finally there are some results of a spatial analysis of R&D spillovers data (for a panel dataset with 22 countries and 20 years) shown. (author's abstract) / Series: Research Report Series / Department of Statistics and Mathematics

Entre as técnicas de análise exploratória de dados espaciais encontram-se os indicadores de associação espacial, que mensuram o grau de dependência espacial dos dados analisados e são aplicáveis apenas a dados quantitativos. Outro procedimento disponível é a geoestatística, a qual se baseia no variograma, descrevendo quantitativa e qualitativamente a estrutura espacial de determinada variável. Neste trabalho, utilizam-se conceitos do variograma para desenvolver um indicador de associação espacial global (SIVAR-G) e um indicador de associação espacial local (SIVAR-L). São utilizados dois bancos de dados: dados binários de escolha modal de uma Cidade Fictícia e dados de média de viagens por modo automóvel por domicílio para um recorte da região central da cidade de São Paulo (Pesquisa de Mobilidade 2012). Em ambos os casos, o indicador global, para diferentes vizinhanças, foi calculado com base em valores padronizados, provenientes do variograma experimental e teórico. Em seguida, aplicou-se um teste de hipótese baseado em pseudo-significância para avaliar a significância do indicador proposto previamente. Por fim, os resultados do indicador proposto foram comparados ao índice de Moran, calculado com os mesmos parâmetros. Para o indicador local, foi elaborado um procedimento similar, porém os cálculos foram feitos pontualmente. Cada observação do banco de dados teve um variograma experimental calculado e um variograma teórico modelado para uma análise omniderecional. Um teste de hipótese similar ao do indicador global foi desenvolvido e aplicado. Assim obtiveram-se indicadores de associação espacial local ponto a ponto. Conclui-se que o indicador SIVAR-G possui desempenho satisfatório na estimação de associação espacial para dados contínuos e binários, mostrando-se sensível a anisotropia dos dados. O indicador SIVAR-L é capaz de identificar \"bolsões\" de associação espacial. É aplicável a dados contínuos e binários. Os indicadores propostos permitem a modelagem de variogramas teóricos globais e locais, fornecendo uma maior riqueza de detalhes da estrutura espacial dos dados. Os indicadores SIVAR baseiam-se na dissimilaridade espacial, enquanto o índice de Moran e LISA baseiam-se na similaridade espacial. / Among the exploratory spatial data analysis tools, there are the indicators of spatial association, which measure the degree of spatial dependence of the analyzed data and can be applied to quantitative data. Other procedure available is the geoestatistics, which is based on the variogram, describing quantitatively and qualitatively the spatial structure of a variable. The aim of this thesis is to use the concept of the variogram to develop a global indicator of spatial association (SIVAR-G) and a local indicator of spatial association (SIVAR-L). Two data bases were used: binary data of travel mode choice of a hypothetical city and mean of automobile trips by household to a region of São Paulo\'s center (Mobility Survey, 2012). In both cases, the global indicator, for different neighborhoods, was calculated based on standardized values, derived from the experimental and theoretical variogram. Then, a pseudo-significance test was applied to evaluate the significance of the previously proposed indicator. The results of the proposed indicator were compared to Moran\'s I, calculated with the same parameters. For the local indicator, it was made a similar procedure, however the calculation was punctual. For each observation of the database, it was calculated experimental variogram and adjusted a theoretical variogram for a omnidirectional analysis. A hypothesis test similar to the one applied in the global indicator was developed and applied. Therefore, it was obtained local indicators point by point. It was concluded that the indicator SIVAR-G has a satisfactory performance for spatial association of binary and continuous data, with sensibility for anisotropy cases. The SIVAR-L indicator is able to identify spatial association pockets and outliers. The local indicator is suitable for continuous and binary data. The developed indicators allow the modeling of theoretical global and local variograms, providing more details of the spatial structure of the data. The SIVAR indicators are based on spatial dissimilarity, while the Moran and LISA index are based on spatial similarity.

Entre as técnicas de análise exploratória de dados espaciais encontram-se os indicadores de associação espacial, que mensuram o grau de dependência espacial dos dados analisados e são aplicáveis apenas a dados quantitativos. Outro procedimento disponível é a geoestatística, a qual se baseia no variograma, descrevendo quantitativa e qualitativamente a estrutura espacial de determinada variável. Neste trabalho, utilizam-se conceitos do variograma para desenvolver um indicador de associação espacial global (SIVAR-G) e um indicador de associação espacial local (SIVAR-L). São utilizados dois bancos de dados: dados binários de escolha modal de uma Cidade Fictícia e dados de média de viagens por modo automóvel por domicílio para um recorte da região central da cidade de São Paulo (Pesquisa de Mobilidade 2012). Em ambos os casos, o indicador global, para diferentes vizinhanças, foi calculado com base em valores padronizados, provenientes do variograma experimental e teórico. Em seguida, aplicou-se um teste de hipótese baseado em pseudo-significância para avaliar a significância do indicador proposto previamente. Por fim, os resultados do indicador proposto foram comparados ao índice de Moran, calculado com os mesmos parâmetros. Para o indicador local, foi elaborado um procedimento similar, porém os cálculos foram feitos pontualmente. Cada observação do banco de dados teve um variograma experimental calculado e um variograma teórico modelado para uma análise omniderecional. Um teste de hipótese similar ao do indicador global foi desenvolvido e aplicado. Assim obtiveram-se indicadores de associação espacial local ponto a ponto. Conclui-se que o indicador SIVAR-G possui desempenho satisfatório na estimação de associação espacial para dados contínuos e binários, mostrando-se sensível a anisotropia dos dados. O indicador SIVAR-L é capaz de identificar \"bolsões\" de associação espacial. É aplicável a dados contínuos e binários. Os indicadores propostos permitem a modelagem de variogramas teóricos globais e locais, fornecendo uma maior riqueza de detalhes da estrutura espacial dos dados. Os indicadores SIVAR baseiam-se na dissimilaridade espacial, enquanto o índice de Moran e LISA baseiam-se na similaridade espacial. / Among the exploratory spatial data analysis tools, there are the indicators of spatial association, which measure the degree of spatial dependence of the analyzed data and can be applied to quantitative data. Other procedure available is the geoestatistics, which is based on the variogram, describing quantitatively and qualitatively the spatial structure of a variable. The aim of this thesis is to use the concept of the variogram to develop a global indicator of spatial association (SIVAR-G) and a local indicator of spatial association (SIVAR-L). Two data bases were used: binary data of travel mode choice of a hypothetical city and mean of automobile trips by household to a region of São Paulo\'s center (Mobility Survey, 2012). In both cases, the global indicator, for different neighborhoods, was calculated based on standardized values, derived from the experimental and theoretical variogram. Then, a pseudo-significance test was applied to evaluate the significance of the previously proposed indicator. The results of the proposed indicator were compared to Moran\'s I, calculated with the same parameters. For the local indicator, it was made a similar procedure, however the calculation was punctual. For each observation of the database, it was calculated experimental variogram and adjusted a theoretical variogram for a omnidirectional analysis. A hypothesis test similar to the one applied in the global indicator was developed and applied. Therefore, it was obtained local indicators point by point. It was concluded that the indicator SIVAR-G has a satisfactory performance for spatial association of binary and continuous data, with sensibility for anisotropy cases. The SIVAR-L indicator is able to identify spatial association pockets and outliers. The local indicator is suitable for continuous and binary data. The developed indicators allow the modeling of theoretical global and local variograms, providing more details of the spatial structure of the data. The SIVAR indicators are based on spatial dissimilarity, while the Moran and LISA index are based on spatial similarity.

The research conducted in this thesis consists of a Geographic Information Systems (GIS) based safety study and a spatial analysis of vehicle crashes in the State of Florida. The GIS safety study is comprised of a County and Roadway Level GIS analysis of multilane corridors. The spatial analysis investigated the use of county-level vehicle crash models, taking spatial effects into account. The GIS safety study examines the locations of high trends of severe crashes (includes incapacitating and fatal crashes) on multilane corridors in the State of Florida at two levels, county level and roadway level. The GIS tool, which is used frequently in traffic safety research, was utilized to visually display those locations. At the county level, several maps of crash trends were generated. It was found that counties with high population and large metropolitan areas tend to have more crash occurrences. It was also found that the most severe crashes occurred in counties with more urban than rural roads. The neighboring counties of Pasco, Pinellas and Hillsborough had high severe crash rate per mile. At the roadway level, seven counties were chosen for the analysis based on their high severe crash trends, metropolitan size and geographical location. Several GIS maps displaying the safety level of multilane corridors in the seven counties were generated. The GIS maps were based on a ranking methodology that was developed in research that evaluated the safety condition of road segments and signalized intersections separately. The GIS maps were supported by Excel tables which provided details on the most hazardous locations on the roadways. The results of the roadway level analysis found that the worst corridors were located in Pasco, Pinellas and Hillsborough Counties. Also, a sliding window approach was developed and performed on the ten most hazardous corridors of the seven counties. The results were graphs locating the most dangerous 0.5 miles on a corridor. For the spatial analysis of crashes, the exploratory Moran's I statistic test revealed that crash related spatial clustering existed at the county level. For crash modeling, a full Bayesian (FB) hierarchical model is proposed to account for the possible spatial correlation among crash occurrence of adjacent counties. The spatial correlation is realized by specifying a Conditional Auto-regressive prior to the residual term of the link function in standard Poisson regression. Two FB models were developed, one for total crashes and one for severe crashes. The variables used include traffic related factors and socio-economic factors. Counties with higher road congestion levels, higher densities of arterials and intersections, higher percentage of population in the 15-24 age group and higher income levels have increased crash risk. Road congestion and higher education levels, however, were negatively correlated with the risk of severe crashes. The analysis revealed that crash related spatial correlation existed among the counties. The FB models were found to fit the data better than traditional methods such as Negative Binomial and that is primarily due to the existence of spatial correlation. Overall, this study provides the Transportation Agencies with specific information on where improvements must be implemented to have better safety conditions on the roads of Florida. The study also proves that neighboring counties are more likely to have similar crash trends than the more distant ones.

The co-occurrence of polyploidy and clonal reproduction among plant species has long been recognized, but the evolutionary mechanisms underlying the association are unknown. Here, I investigate whether polyploidy increases the magnitude of clonality, either directly or indirectly, by comparing the extent and spatial structure of clones between diploid and tetraploid Chamerion angustifolium in a greenhouse environment and natural populations. In the greenhouse, tetraploid plants allocated 90.4% more dry mass to root buds, the primary mechanism of clonal reproduction, than diploids. Per unit root mass, tetraploids produced 44% fewer root buds and the average position of the root buds along the root was 47% closer to the stem than in diploids. In natural populations, the magnitude of clonality in tetraploid C. angustifolium was similar or less than in diploids. However, clones were spatially aggregated in all diploid populations but only in two of five tetraploid populations. Average clone patch diameter, however, was not significantly different between diploids (3.9 m) and tetraploids (2.5 m). These data do not support the hypothesis that clonality increases as a result of genome duplication. Rather, it is possible that clonality is linked to genome duplication because clonal diploids are predisposed for polyploid formation and establishment. / National Science and Engineering Research Council, Canada Research Chair Program, Canadian Foundation for Innovation

No description available.

Background: Adverse birth outcomes present a difficult and chronic challenge in Ontario, in Canada and in developed countries in general. Increasing proportions of preterm births, significant regional disparities and the high cost of treating all adverse birth outcomes have focused attention on explaining them and developing effective treatments. Methods: Birth outcomes and maternal characteristics for approximately 626,000 births, about 90% of births in 2005–2009, were linked to small geographic areas throughout Ontario. For each of four adverse outcomes: late preterm, moderate to very preterm, small for gestation age and still births, proportions of total births were calculated for the full province and for each small geographic area. Geographic hotspots of elevated rates were identified for each of the different adverse birth outcomes using the local Moran’s I statistic. Data for nine known ecologic and individual risk factors were then linked to the areas. Hierarchical regression analysis was used to model each of the outcomes for the full province and for dispersed local areas. The resulting models for the different outcomes were contrasted. Results: Significant geographic hotspots exist for each of the four outcomes. Hotspots for the different outcomes were found to be largely spatially exclusive. For like outcomes, predictive models differed markedly between local areas (i.e. local groups of hotspots) as well as between full-province and local areas. Ecologic level variables played a strong role in all models; the influence of individual level risk factors was consistently modified by ecologic risk factors except for small for gestational births. Conclusions: The finding of significant hotspots for different adverse birth outcomes indicates that certain geographic areas have aetiologies or patterns of predictors sufficient to create significantly elevated levels of particular outcomes. The finding that hotspots for the different adverse outcomes are largely exclusive implies that the aetiologies are specific; i.e., those that are sufficient to create significantly higher levels for one outcome do not also create significantly higher levels of others. The consistently strong role of ecologic level risk factors in modifying individual level risk factors implies that contextual characteristics are an important part of the aetiology of adverse birth outcomes. Differences in local area models suggest the existence of location-specific (rather than universal) aetiologies. The findings support the need for more careful attention to local context when explaining birth outcomes.

La compréhension des processus écologiques qui sous-tendent l’assemblage des communautés végétales et la coexistence des espèces est un objectif central en écologie. Ces processus sont potentiellement nombreux et de natures contrastées. Ainsi, la composition d’une communauté de plantes dépend de processus déterministes liés aux conditions environnementales abiotiques (climat, conditions physiques et chimiques du sol, lumière) et d’interactions biotiques complexes, positives (facilitation, symbioses) comme négatives (compétition, prédation, pathogènes). En outre, les communautés sont influencées par des processus stochastiques (capacité de dispersion limitée, dérive écologique). Si les mécanismes à l’origine de ces processus sont très différents, ils ont néanmoins en commun la génération de motifs (patterns) spatiaux de distribution d’espèces dans les communautés. L’analyse de la structure spatiale des communautés permet ainsi une étude indirecte des processus régissant les communautés. La nature complexe de ces patterns spatiaux a mené au développement de nombreuses méthodes statistiques de détection et de description de patterns. Les méthodes basées sur des vecteurs propres spatiaux sont parmi les plus puissantes et précises pour détecter des patterns complexes et multi-échelles. Ces vecteurs propres, utilisés comme prédicteurs spatiaux, peuvent être combinés à un ensemble de variables environnementales dans un cadre de partition de variation. Celui-ci permet, en théorie, de démêler les effets uniques et l’effet conjoint des variables environnementales et spatiales sur la variation de composition d’une communauté. Il mène ainsi à une quantification de l’action des processus déterministes et des processus stochastiques sur l’assemblage de la communauté. Néanmoins, je montre dans cette thèse qu’un certain flou méthodologique concernant deux étapes déterminantes des analyses basées sur les vecteurs propres spatiaux a mené une proportion élevée d’études à utiliser ces méthodes de manière sous-optimale, voire fortement biaisée. Ceci compromet la fiabilité des patterns spatiaux détectés et des processus écologiques inférés. Une autre limitation de ce cadre d’analyse concerne la fraction de la partition de variation décrivant l’effet environnemental spatialement structurés qu’aucune méthode ne permet de tester.Cette thèse présente des solutions non biaisées, puissantes et précises à ces différentes limitations méthodologiques et permet d’élargir le cadre de l’inférence de processus écologique à partir de patterns spatiaux de communautés. Les différentes étapes d’amélioration de ces méthodes ont également été illustrées dans la thèse au travers de trois cas d’études fournis par deux communautés d’arbres tropicale et tempérée et une communauté de champignons symbiotiques des arbres. / Doctorat en Sciences / info:eu-repo/semantics/nonPublished