Made available in DSpace on 2017-07-10T19:24:36Z (GMT). No. of bitstreams: 1 Mario Antonio Faraco.pdf: 2332700 bytes, checksum: 571424bd7076023344d5d704ea9ab256 (MD5) Previous issue date: 2006-05-03 / Researches about the spatial variability of the soil attributes that influence the productivity are highly important for the development of new technologies that benefits the agriculture. To verify the variability of these attributes we used the geostatistic that offers techniques to the obtainment of information concerning this variability. The processes of data analysis use methods that include optimization algorithms to the choice a geostatistic model and the estimation of that model parameters. We studied the following soil attributes: soil resistance to penetration, soil density, soil humidity and the soybean s productivity variable. This paper has as its purpose to describe the spatial behavior of empiric and simulated data and to build models of spatial variability to the attributes in study with the main purpose of evaluating the quality of the adjustments according to the Criterions of Akaike, Filliben, Jackknifing and Cross Validation. The research was developed in the West Parana region, in a area of 54 ha where the typical soil of the region is the Red Distrofic Latosoil and a net with a 100 georeferred parcels was utilized. To the structure analyze of spatial dependency we used experimental semivariograms generated from sample data set. Afterward those theoretical models were adjusted to the experimental semivariograms and techniques of evaluation of the adjustments were applied to the geoestatistic models. Consequently, the results of the several methods studied were analyzed and the gotten results were compared, concluding for the cross validation as the best adjustment criterion. / Pesquisas sobre a variabilidade espacial dos atributos do solo que influenciam a produtividade são de suma importância para o desenvolvimento de novas tecnologias que beneficiam a agricultura. Para verificar a variabilidade desses atributos utilizou-se a geoestatística que disponibiliza técnicas para a obtenção de informações a respeito dessa variabilidade. Os processos de análise de dados utilizam métodos que incluem algoritmos de otimização para escolha de um modelo geoestatístico e a estimação dos parâmetros desse modelo. Foram estudados os atributos do solo: resistência do solo à penetração, densidade do solo, umidade do solo e a variável produtividade da soja. Este trabalho tem por objetivo descrever os comportamentos espaciais de dados empíricos e simulados e construir modelos de variabilidade espacial para os atributos em estudo com o objetivo principal de avaliar a qualidade dos ajustes segundo os Critérios de Akaike, Filliben, Jackknifing e Validação Cruzada. A pesquisa foi desenvolvida na região Oeste do Paraná, em uma área de 57 há, cujo solo típico é o Latossolo Vermelho Distrófico e foi utilizada uma malha com 100 parcelas georeferenciadas. Para a análise da estrutura de dependência espacial foram utilizados emivariogramas experimentais gerados a partir dos dados amostrais. Em seguida, ajustaram-se modelos teóricos aos emivariogramas experimentais e aplicaram-se as técnicas de avaliação dos ajustes aos modelos geoestatísticos. Em conseqüência, analisaram-se os resultados dos diversos métodos estudados, compararando-se os resultados obtidos e concluindo-se pela validação cruzada como o melhor critério de ajuste.

Respondent-driven sampling (RDS) is very useful in collecting data from individuals in hidden populations, where a sampling frame does not exist. It starts with researchers choosing initial respondents from a group which may be involved in taboo or illegal activities, after which they recruit other peers who belong to the same group. Analysis results in unbiased estimates of population proportions though with strong assumptions about the underlying social network and RDS recruitment process. These assumptions bear little resemblance to reality, and thus compromise the estimation of any means, population proportions or variances inferred from studies. The topology of the contact network, denoted by the number of links each person has, provides insight into the processes of infectious disease spread. The overall objective of the thesis is to identify the topology of an injection drug use network, and critically review the methods developed to produce estimates. The topology of people who inject drugs (PWID) collected by RDS in Ottawa, 2006 was compared with a Poisson distribution, an exponential distribution, a power-law distribution, and a lognormal distribution. The contact distribution was then evaluated against a small-world network characterized by high clustering and low average distances between individuals. Last a systematic review of the methods used to produce RDS mean and variance estimates was conducted. The Poisson distribution, a type of random distribution, was not an appropriate fit for PWID network. However, the PWID network can be classified as a small world network organised with many connections and short distances between people. Prevention of transmission in such networks should be focussed on the most active people (clustered individuals and hubs) as intervention with any others is less effective. The systematic review contained 32 articles which included the development and evaluation of 12 RDS mean and 6 variance estimators. Overall, the majority of estimators perform roughly the same, with the exception of RDSIEGO which outperformed the 6 other RDS mean estimators. The Tree bootstrap variance estimate does not rely on modelling RDS as a first order Markov (FOM) process, which seems to be the main limitation of the other existing estimators. The lack of FOM as an assumption and the flexible application of this variance estimator to any RDS point estimate make the Tree bootstrapping estimator a more efficient choice.

Made available in DSpace on 2017-05-12T14:47:59Z (GMT). No. of bitstreams: 1 Mario Antonio Faraco.pdf: 2332700 bytes, checksum: 571424bd7076023344d5d704ea9ab256 (MD5) Previous issue date: 2006-05-03 / Researches about the spatial variability of the soil attributes that influence the productivity are highly important for the development of new technologies that benefits the agriculture. To verify the variability of these attributes we used the geostatistic that offers techniques to the obtainment of information concerning this variability. The processes of data analysis use methods that include optimization algorithms to the choice a geostatistic model and the estimation of that model parameters. We studied the following soil attributes: soil resistance to penetration, soil density, soil humidity and the soybean s productivity variable. This paper has as its purpose to describe the spatial behavior of empiric and simulated data and to build models of spatial variability to the attributes in study with the main purpose of evaluating the quality of the adjustments according to the Criterions of Akaike, Filliben, Jackknifing and Cross Validation. The research was developed in the West Parana region, in a area of 54 ha where the typical soil of the region is the Red Distrofic Latosoil and a net with a 100 georeferred parcels was utilized. To the structure analyze of spatial dependency we used experimental semivariograms generated from sample data set. Afterward those theoretical models were adjusted to the experimental semivariograms and techniques of evaluation of the adjustments were applied to the geoestatistic models. Consequently, the results of the several methods studied were analyzed and the gotten results were compared, concluding for the cross validation as the best adjustment criterion. / Pesquisas sobre a variabilidade espacial dos atributos do solo que influenciam a produtividade são de suma importância para o desenvolvimento de novas tecnologias que beneficiam a agricultura. Para verificar a variabilidade desses atributos utilizou-se a geoestatística que disponibiliza técnicas para a obtenção de informações a respeito dessa variabilidade. Os processos de análise de dados utilizam métodos que incluem algoritmos de otimização para escolha de um modelo geoestatístico e a estimação dos parâmetros desse modelo. Foram estudados os atributos do solo: resistência do solo à penetração, densidade do solo, umidade do solo e a variável produtividade da soja. Este trabalho tem por objetivo descrever os comportamentos espaciais de dados empíricos e simulados e construir modelos de variabilidade espacial para os atributos em estudo com o objetivo principal de avaliar a qualidade dos ajustes segundo os Critérios de Akaike, Filliben, Jackknifing e Validação Cruzada. A pesquisa foi desenvolvida na região Oeste do Paraná, em uma área de 57 há, cujo solo típico é o Latossolo Vermelho Distrófico e foi utilizada uma malha com 100 parcelas georeferenciadas. Para a análise da estrutura de dependência espacial foram utilizados emivariogramas experimentais gerados a partir dos dados amostrais. Em seguida, ajustaram-se modelos teóricos aos emivariogramas experimentais e aplicaram-se as técnicas de avaliação dos ajustes aos modelos geoestatísticos. Em conseqüência, analisaram-se os resultados dos diversos métodos estudados, compararando-se os resultados obtidos e concluindo-se pela validação cruzada como o melhor critério de ajuste.

In this work we address the problem of friction compensation in a pneumatic control valve. It is proposed a nonlinear control law that uses algebraic estimators in its structure, in order to adapt the controller to the aging of the valve. For that purpose we estimate parameters related to the valve\'s Karnopp model, necessary to friction compensation, online. The estimators and the controller are validated through simulations. / Nessa pesquisa, estudamos o problema de compensação de atrito em válvulas pneumáticas. É proposta uma lei de controle não linear que tem estimadores algébricos em sua estrutura, para adaptar o controlador ao envelhecimento da válvula. Para isso, estimam-se os valores de parâmetros relacionados ao modelo de Karnopp da válvula, necessários à compensação do atrito, de maneira online. Os estimadores e o controlador são validados através de simulações.

The thesis describes Random Coefficient Autoregressive time series mo- dels (RCA models). In first chapter we introduce different types of estimati- ons for coefficients of RCA model. Main part is in second chapter, where we describe detection changes procedures for all methods mentioned in chapter one, here the thesis expands the current theory about change detection of wei- ghted least square method and functional estimation. In last chapter we sum- marize results of simulation study. 1

The purpose of this research is to investigate the performance of some ridge regression estimators for the logistic regression model in the presence of moderate to high correlation among the explanatory variables. As a performance criterion, we use the mean square error (MSE), the mean absolute percentage error (MAPE), the magnitude of bias, and the percentage of times the ridge regression estimator produces a higher MSE than the maximum likelihood estimator. A Monto Carlo simulation study has been executed to compare the performance of the ridge regression estimators under different experimental conditions. The degree of correlation, sample size, number of independent variables, and log odds ratio has been varied in the design of experiment. Simulation results show that under certain conditions, the ridge regression estimators outperform the maximum likelihood estimator. Moreover, an empirical data analysis supports the main findings of this study. This thesis proposed and recommended some good ridge regression estimators of the logistic regression model for the practitioners in the field of health, physical and social sciences.

(1) Reliable measures of population abundance are essential for managing wildlife effectively. Aerial surveys provide a rapid and efficient means of surveying large mammals and many techniques have been developed to adjust for the inability to count all animals within transects. The probability of detection varies according to a range of factors which are important to consider when estimating density. Standardised survey methods developed in flat country are not readily transferable to steep terrain due to safety, access and difficulties delineating transect widths. Other methods have logistic constraints and must adhere to various other assumptions. (2) Density estimators are seldom examined using actual population size, hence their ability to correct for true bias is unknown. Studies that compare techniques are difficult to interpret because of the uncertainty of adherence to their respective assumptions. Factors influencing detection probability, estimators that correct for bias, the validity of their assumptions and how these relate to true density are important considerations for selecting suitable methods. The aim of this study was to obtain accurate and reliable methods for estimating the density of feral goats by improving predictions of detection probability, investigating the assumptions of aerial surveys, and examining the accuracy of 15 density estimators by comparing with total counts of feral goats. (3) Group size, vegetation and observer were the most important factors influencing the probability of observing a group of goats during aerial surveys. However, different approaches to analysing these data influenced the significance of variables and the predicted probabilities. Goat colour, type of helicopter, site and rear observer experience in hours were also found to be significant (P<0.05) when using likelihood equations based on all animals in the population rather than only those in the sample. The slope of the terrain was also shown to significantly (P=0.014) affect the probability of detection. (4) Indices are commonly used in wildlife management for their simplicity and practicality, but their validity has been questioned because of variable probability of detection. Results of this study suggest aerial survey indices are useful in monitoring a range of medium-sized mammal species across space and time if differences in detection probability between species, group size, vegetation and observer are considered and their effects are standardised. (5) An assumption of most sampling regimes that is fundamental but rarely examined is that animals are not counted more than once. In this study the behavioural responses of feral goats to helicopters were investigated as a basis for estimating the probability that goats were recounted. No long-term consequences were evident in feral goat behaviour of responses to helicopters. However, helicopter surveys were found to alter the structure of 42% of groups observed, with 28% of groups merging with others and 14% splitting into separate groups. Therefore, group size estimated from the air should not be considered as biologically important, and when estimating density, researchers should also avoid using group sizes determined from independent ground observations to correct group sizes determined from aerial surveys. Goats were also more likely to flush further when helicopters were within 150 m, which is close to or within standard helicopter strip widths. Substantial movement occurred between transects and 21% of goats were estimated to be available for recounting in adjacent transects. (6) Different detection probabilities between groups of goats may be particularly relevant when using double-counting, where multiple observers are �capturing� and �recapturing� animals in the same instant. Many analyses test and adjust for this �unequal catchability� assumption in different ways, with the approaches of Huggins and Alho allowing prediction of unique probability values for a range of co-variates. The approach of Chao attempts to correct for skewed distributions in small samples. The Horvitz-Thompson approach provides a useful basis for estimating abundance (or density) when detection probability can be estimated and is known to vary between observations according to a range of independent variables, and also avoids errors associated with averaging group size. (7) After correcting for recounting, the Alho estimator applied to helicopter surveys was the most accurate (Bias = 0.02) and reliable of all techniques, which suggests that estimates were improved by taking into account unconditional detection probability and correcting individual observations according to their characteristics. The positive bias evident in the Chao (Bias = 0.28) and Petersen (Bias = 0.15) aerial survey estimators may have been a result of averaging detection probability across all observations. The inconsistency and inaccuracy of the ground-based area-count technique emphasises the importance of other assumptions in density estimation, such as representative sampling and availability bias. The accuracy of index-manipulation-index techniques was dependent on the indices used. Capture-recapture estimates using mustering showed slight negative bias (Bias = -0.08), which was likely a result of increased probability of re-capture (i.e. trap happy). Ground-based capture-resight estimates were labour intensive and positively biased (Bias = 0.13), likely due to underestimating the area sampled, or overestimating the number of unmarked individuals with each sample. (8) Helicopter survey using double-counting is recommended for estimating the density of feral goats in steep terrain. However, consideration of recounting under intensive sampling regimes and adjustments for the factors that influence unconditional detection probability is required.

In a nonparametric setting, the functional form of the relationship between the response variable and the associated predictor variables is unspecified; however it is assumed to be a smooth function. The main aim of nonparametric regression is to highlight an important structure in data without any assumptions about the shape of an underlying regression function. In regression, the random and fixed design models should be distinguished. Among the variety of nonparametric regression estimators currently in use, kernel type estimators are most popular. Kernel type estimators provide a flexible class of nonparametric procedures by estimating unknown function as a weighted average using a kernel function. The bandwidth which determines the influence of the kernel has to be adapted to any kernel type estimator. Our focus is on Nadaraya-Watson estimator and Local Linear estimator which belong to a class of kernel type regression estimators called local polynomial kerne l estimators. A closely related problem is the determination of an appropriate sample size that would be required to achieve a desired confidence level of accuracy for the nonparametric regression estimators. Since sequential procedures allow an experimenter to make decisions based on the smallest number of observations without compromising accuracy, application of sequential procedures to a nonparametric regression model at a given point or series of points is considered. The motivation for using such procedures is: in many applications the quality of estimating an underlying regression function in a controlled experiment is paramount; thus, it is reasonable to invoke a sequential procedure of estimation that chooses a sample size based on recorded observations that guarantees a preassigned accuracy. We have employed sequential techniques to develop a procedure for constructing a fixed-width confidence interval for the predicted value at a specific point of the independent variable. These fixed-width confidence intervals are developed using asymptotic properties of both Nadaraya-Watson and local linear kernel estimators of nonparametric kernel regression with data-driven bandwidths and studied for both fixed and random design contexts. The sample sizes for a preset confidence coefficient are optimized using sequential procedures, namely two-stage procedure, modified two-stage procedure and purely sequential procedure. The proposed methodology is first tested by employing a large-scale simulation study. The performance of each kernel estimation method is assessed by comparing their coverage accuracy with corresponding preset confidence coefficients, proximity of computed sample sizes match up to optimal sample sizes and contrasting the estimated values obtained from the two nonparametric methods with act ual values at given series of design points of interest. We also employed the symmetric bootstrap method which is considered as an alternative method of estimating properties of unknown distributions. Resampling is done from a suitably estimated residual distribution and utilizes the percentiles of the approximate distribution to construct confidence intervals for the curve at a set of given design points. A methodology is developed for determining whether it is advantageous to use the symmetric bootstrap method to reduce the extent of oversampling that is normally known to plague Stein's two-stage sequential procedure. The procedure developed is validated using an extensive simulation study and we also explore the asymptotic properties of the relevant estimators. Finally, application of our proposed sequential nonparametric kernel regression methods are made to some problems in software reliability and finance.

In this thesis, pairwise multiple comparisons and multiple comparisons with a control are studied when the observations have short-tailed symmetric distributions. Under non-normality, the testing procedure is given and Huber estimators, trimmed mean with winsorized standard deviation, modified maximum likelihood estimators and ordinary sample mean and sample variance used in this procedure are reviewed. Finally, robustness properties of the stated estimators are compared with each other and it is shown that the test based on the modified maximum likelihood estimators has better robustness properties under short-tailed symmetric distribution.

For statistical estimation of population parameters, Fisher&rsquo / s maximum likelihood estimators (MLEs) are commonly used. They are consistent, unbiased and efficient, at any rate for large n. In most situations, however, MLEs are elusive because of computational difficulties. To alleviate these difficulties, Tiku&rsquo / s modified maximum likelihood estimators (MMLEs) are used. They are explicit functions of sample observations and easy to compute. They are asymptotically equivalent to MLEs and, for small n, are equally efficient. Moreover, MLEs and MMLEs are numerically very close to one another. For calculating MLEs and MMLEs, the functional form of the underlying distribution has to be known. For machine data processing, however, such is not the case. Instead, what is reasonable to assume for machine data processing is that the underlying distribution is a member of a broad class of distributions. Huber assumed that the underlying distribution is long-tailed symmetric and developed the so called M-estimators. It is very desirable for an estimator to be robust and have bounded influence function. M-estimators, however, implicitly censor certain sample observations which most practitioners do not appreciate. Tiku and Surucu suggested a modification to Tiku&rsquo / s MMLEs. The new MMLEs are robust and have bounded influence functions. In fact, these new estimators are overall more efficient than M-estimators for long-tailed symmetric distributions. In this thesis, we have proposed a new modification to MMLEs. The resulting estimators are robust and have bounded influence functions. We have also shown that they can be used not only for long-tailed symmetric distributions but for skew distributions as well. We have used the proposed modification in the context of experimental design and linear regression. We have shown that the resulting estimators and the hypothesis testing procedures based on them are indeed superior to earlier such estimators and tests.