This dissertation comprises of three original contributions to empirical forecasting research. Chapter 1 introduces the dissertation. Chapter 2 contributes to the literature on hierarchical time series (HTS) modelling by proposing a disaggregated forecasting system for both inflation rate and its volatility. Using monthly data that underlies the Retail Prices Index for the UK, we analyse the dynamics of the inflation process. We examine patterns in the time-varying covariation among product-level inflation rates that aggregate up to industry-level inflation rates that in turn aggregate up to the overall inflation rate. The aggregate inflation volatility closely tracks the time path of this covariation, which is seen to be driven primarily by the variances of common shocks shared by all products, and by the covariances between idiosyncratic product-level shocks. We formulate a forecasting system that comprises of models for mean inflation rate and its variance, and exploit the index structure of the aggregate inflation rate using the HTS framework. Using a dynamic model selection approach to forecasting, we obtain forecasts that are between 9 and 155 % more accurate than a SARIMA-GARCH(1,1) for the aggregate inflation volatility. Chapter 3 is on improving forecasts using forecast combinations. The paper documents the software implementation of the open source R package for forecast combination that we coded and published on the official R package depository, CRAN. The GeomComb package is the only R package that covers a wide range of different popular forecast combination methods. We implement techniques from 3 broad categories: (a) simple non-parametric methods, (b) regression-based methods, and (c) geometric (eigenvector) methods, allowing for static or dynamic estimation of each approach. Using S3 classes/methods in R, the package provides a user-friendly environment for applied forecasting, implementing solutions for typical issues related to forecast combination (multicollinearity, missing values, etc.), criterion-based optimisation for several parametric methods, and post-fit functions to rationalise and visualise estimation results. The package has been listed in the official R Task Views for Time Series Analysis and for Official Statistics. The brief empirical application in the paper illustrates the package’s functionality by estimating forecast combination techniques for monthly UK electricity supply. Chapter 4 introduces HTS forecasting and forecast combination to a healthcare staffing context. A slowdown of healthcare budget growth in the UK that does not keep pace with growth of demand for hospital services made efficient cost planning increasingly crucial for hospitals, in particular for staff which accounts for more than half of hospitals’ expenses. This is facilitated by accurate forecasts of patient census and churn. Using a dataset of more than 3 million observations from a large UK hospital, we show how HTS forecasting can improve forecast accuracy by using information at different levels of the hospital hierarchy (aggregate, emergency/electives, divisions, specialties), compared to the naïve benchmark: the seasonal random walk model applied to the aggregate. We show that forecast combination can improve accuracy even more in some cases, and leads to lower forecast error variance (decreasing forecasting risk). We propose a comprehensive parametric approach to use forecasts in a nurse staffing model that has the aim of minimising cost while satisfying that the care requirements (e.g. nurse hours per patient day thresholds) are met.

Diploma thesis extends my Bachelor´s thesis, which was about methods of time distribution of precipitations and differences in that distribution within a small area. The first part of the paper describes the formation and distribution of precipitation, typical variation of precipitation and some characteristics, revealing and describing the behaviour of rainfall in a given station. Following part describes possible methods of evaluation of time distribution. Statistical methods are presented mainly here. Here described methods and evaluations are subsequently applied in time series of daily precipitations, which has been available for this work for the Jenin creek basin. The analysis is focused on characteristics and evaluation of rainfall between terms 1980 - 1988 and 2005 - 2013. The results show us some differences between terms in rainfall characteristics. However, mostly they are not statistically significant. The work also evaluates differences in rainfall totals between rainfall stations Jenín and Dolní Dvořiště (approximate distance of the stations is 4 km) in 2005 - 2013 term. During the evaluation it comes to differences in rainfall totals. They are highly variable and they don´t have neither constant direction nor size.

The aim of the PhD dissertation is to research political trust, specifically to describe trends in political trust and to identify factors which are related to reported trust. Political trust is vertical trust oriented to selected political institutions which are grounded in the Constitution of the Czech Republic. Following political institutions are analyzed in this study: the Government of the Czech Republic, the Chamber of Deputies, the Senate, regional councils and municipal councils. Life-time learning model proposed by Mishler and Rose, which combines cultural and institutional approaches to explanation of trust in political institutions, is used as a theoretical framework in this study. All conducted analyses make use of data from surveys carried out by CVVM and other departments of the Institute of Sociology of the Academy of Sciences of the Czech Republic. The key finding of the study is the fact that the level of trust in political institutions is mainly determined by factors associated with 1) institutional performance which is related to political situation and the perception of own economic situation or the economic situation of the state; 2) party preferences (presence of so called "the winner effect"). The level of trust in political institutions on national, regional, and local...

x, 57 p. : ill. (mostly col.) / An extensive area west of the Three Sisters volcanoes of Oregon has been actively uplifting for over a decade. Examining the deformation is imperative to improve understanding of the potential hazards of Cascade volcanism and the emplacement of magma. I refine the timing of the onset of the deformation, resolve the change in uplift rates through time, and quantify the current deformation rate using Interferometric Synthetic Aperture Radar. The deformation is assessed in time and space using single interferogram InSAR, stacks of interferograms, and line-of-sight time series. I examine the shape of the deformation pattern and explore volcanic source parameters using a Mogi model and tension crack model with topographic corrections. By using the best fit model and combining all useable interferograms from different tracks, I create the first complete continuous inflation time series of the Three Sisters volcanic uplift from 1992 to 2010. / Committee in charge: Dr. David A. Schmidt, Chair; Dr. Katharine V. Cashman, Member; Dr. Joshua J. Roering, Member

The cycle regression analysis algorithm is the most recent addition to a group of techniques developed to detect "hidden periodicities." This dissertation investigates four major aspects of the algorithm. The objectives of this research are 1. To develop an objective method of obtaining an initial estimate of the cycle period? the present procedure of obtaining this estimate involves considerable subjective judgment; 2. To validate the algorithm's success in extracting cycles from multi-cylical data; 3. To determine if a consistent relationship exists among the smallest amplitude, the error standard deviation, and the number of replications of a cycle contained in the data; 4. To investigate the behavior of the algorithm in the predictions of major drops.

During the past 30-years, Biochemical System Theory (BST) has been provided a concrete foundation for the study of the dynamic biological systems, e.g. S-systems models for reverse engineering of metabolic networks (Savageau, 1969; Savageau, 1970; Voit, 2000). One of the remarkable characteristics of these models is its parameters not only quantify the interactions between the components of the network, but also elucidate the networks topology. Automatic procedures for S-system parameterization from biological time series have been developed by many researches, where they assume a noise-free time series and a true estimated first derivative in their methodologies (Chou, et al., 2006; Kikuchi, et al., 2003). Nevertheless, this noise-free data is not a realistic scenario of the real biological experimental world. Methods as artificial neural network (ANN), Support Vectors Machines (SVM) and Saviztsky-Golay filter were proposed to overcome the denoising time series problem with the advantage of a closed form output which allowed determining the first derivative symbolically (Almeida and Voit, 2003; Borges, et al., 2006; Borges, et al., 2004; Voit and Almeida, 2004). However, these solutions showed some problematic artifacts in its first derivative even when they are not visually apparent in the smoothed data, leaving a gap on the issue of a fully automatic method for S-system parameterization from experimental data. The algorithm presented in this work is a proposal to fill this gap up providing an unbiased robust tool for signal extraction and first derivative estimation from noisy time series. / Nos últimos 30 anos, a Teoria dos Sistemas Bioquímicos (Biochemical System Theory - BST) tem fornecido uma fundação concreta para o estudo da dinâmica de sistemas biológicos, por exemplo, Sistemas-S (S-systems) usados em engenharia reversa de vias metabólicas (Savageau, 1969; Savageau, 1970; Voit, 2000). Uma característica marcante desse tipo de modelo é que os parâmetros não só quantificam as interações entres os componentes da rede metabólica, mas também fornecem a sua topologia de regulação. Procedimentos automáticos para a parametrização dos Sistemas-S a partir de séries temporais biológicas vêm sendo desenvolvidos por vários pesquisadores, onde se assume que a série temporal e sua derivada temporal são livres de ruído. Entretando, perfis metabólicos livres de ruído não são um realistas em cenários de experimentos de biologia molecular. Técnicas como Redes Neurais Artificiais (RNA), Máquinas de Vetores de Suporte (MVP) e filtro de Saviztsky-Golay foram propostas como solução do problema de suavização dos perfis metabólicos com a vantagem da obtenção da derivada temporal simbólica (Almeida and Voit, 2003; Borges, et al., 2006; Borges, et al., 2004; Voit and Almeida, 2004). Entretanto, essas soluções apresentaram alguns artefatos problemáticos na derivada até mesmo quando nenhum problema é visualmente detectado no dado suavizado, deixando aberto um espaço vazio na questão de um método automático para a parametrização dos Sistemas-S a partir de dados experimentais. O algoritmo apresentado neste trabalho propõe preencher esse espaço com uma ferramenta robusta para a extração de sinal e de sua derivada temporal a partir de séries temporais ruidosas.

This study aims to gather, analyze, and interpret the potential and economic feasibility of sequestering carbon dioxide and/or generating renewable energy from the processing of pig manure, via time series analysis, using the Box and Jenkins (1970) models for future prediction. Swine production systems generate large amounts of waste which are highly polluting and impact the environment, principally greenhouse gases. When treated, these gases produce methane gas and in turn biogas, which when removed correctly is converted into electricity and carbon credits, leaving behind a high quality biofertilizer. This scenario of carbon dioxide sequestration was only possible after the adoption of the Kyoto Protocol, which provides for the Clean Development Mechanism (CDM) and the sale of Certified Emission Reductions (CER); electrical energy was affected by Regulatory Resolution 482/ 2012 of the National Electric Energy Agency (ANEEL). Data were was gathered with the support of the Municipal Association of Swine Producers of Toledo (AMST) and the BRF Company. This study demonstrates that it is feasible to use biogas and that the payback period follows a production scale with equipment operating 10 hours or more daily. Consequently, the payback period will be between 70 and 80 months, with production of electrical energy and/or in conjunction with the production of carbon credits. The study also shows that the production volume is feasible and should accommodate a minimum number of pigs. In addition, prediction models are important tools since they anticipate the behavior of the series under study, providing support so that investments can be made, reducing risks to the both the swine producer (in the area of finances) and to the farm, as well as allowing for another source of income. By reducing production costs, autonomy is guaranteed in ensuring an uninterrupted supply of electricity, which is needed for swine operations which become more and more dependent on technology. / O presente estudo visa reunir, analisar e interpretar o potencial e a viabilidade econômica do sequestro de gás carbônico e/ou a geração de energia renovável proveniente do tratamento de dejetos suínos, por meio da análise de séries temporais, com utilização dos modelos de previsão Box e Jenkins (1970). Como os sistemas de produção de suínos geram grande quantidade de dejetos altamente poluentes e impactantes no meio ambiente, principalmente os gases de efeito estufa, estes, ao serem tratados, produzirão o gás metano e dentro deste o biogás que, ao ser eliminado corretamente é convertido em energia elétrica e em créditos de carbono, restando na propriedade um biofertilizante de alta qualidade. Cenário do sequestro de gás carbônico que só foi possível após a aprovação do Protocolo de Quioto que prevê o Mecanismo de Desenvolvimento Limpo (MDL), a comercialização dos Certificados de Emissões Reduzidas (CER) e, no âmbito da energia elétrica, pela Resolução Regulamentadora 482/2012 da Agência Nacional de Energia Elétrica (ANEEL). Para o levantamento dessas informações, contou-se com o apoio da Associação Municipal dos Suinocultores de Toledo (AMST) e da Empresa BRF. Esta pesquisa demonstra que é viável a utilização do biogás e que o período de retorno obedece a uma escala de produção com um funcionamento diário dos equipamentos de 10 horas ou mais. Dessa forma, o período do retorno do investimento será entre 70 a 80 meses, com a produção de energia elétrica e/ou em conjunto com a produção de créditos de carbono. Mostra também que o volume de produção é viável, desde que aloje um número mínimo de suínos. Além disso, o uso de modelos de previsão é importante ferramenta, pois antecipam o comportamento da série em estudo, fornecendo ao produtor subsídios para que o investimento seja feito reduzindo o risco ao suinocultor, tanto nos aspectos financeiros, pois a granja terá mais uma fonte de renda, quanto na redução dos custos de produção, já que terá autonomia no fornecimento de energia elétrica ininterruptamente, o que é necessário nos sistemas de criação cada vez mais tecnificados.

Submitted by RAFAEL LOPES PAIXÃO DA SILVA (lopes1313@gmail.com) on 2018-05-02T19:05:38Z No. of bitstreams: 1 0-Thesis.pdf: 4208376 bytes, checksum: e3d171683a2ab6be4462439595e613c3 (MD5) / Approved for entry into archive by Hellen Sayuri Sato null (hellen@ift.unesp.br) on 2018-05-03T18:00:03Z (GMT) No. of bitstreams: 1 silva_rlp_me_ift.pdf: 4208376 bytes, checksum: e3d171683a2ab6be4462439595e613c3 (MD5) / Made available in DSpace on 2018-05-03T18:00:03Z (GMT). No. of bitstreams: 1 silva_rlp_me_ift.pdf: 4208376 bytes, checksum: e3d171683a2ab6be4462439595e613c3 (MD5) Previous issue date: 2018-04-03 / Conselho Nacional de Desenvolvimento Científico e Tecnológico (CNPq) / A modelagem matemática é uma ferramenta presente nos campos da ecologia teórica e da biologia ma- temática. Porém tais modelos que tentam reproduzir parte da dinâmica natural são limitados, o que rapidamente esgota as possibilidades de investigações e exploração dos dados. Visando contornar isso partimos para o contexto da reconstrução de espaços-de-fase, pois queremos obter outras informações sobre aquilo que temos em mãos, a observação da natureza, o dado. De posse dessa nova aplicação da teoria de sistemas dinâmicos, é nos possibilitado uma nova inferência sobre o fenômeno observado, bem como suas causas que, através do modelo estavam ocultas. A técnica do mapeamento cruzado convergente, entre atratores gerados pela reconstrução de espaços-de-fase, através da representação do espaço-de-fase original num espaço euclidiano formado pela série temporal original e seus atrasos, pos- sibilita uma inferência de causalidade mais pragmática e mais efetiva para sistemas que obedeçam uma dinâmica não-linear, o caso para as muitas séries ecológicas e biológicas de interesse. / Mathematical modeling is an almost omnipresent tool in the fields of theoretical ecology and mathe- matical biology. However, such models that try to partially reproduce the natural dynamics are limited, which quickly runs out possibilities for data-driven investigation and exploration. Aiming to circumvent this, we set out to the context of phase-space reconstruction, since we want to obtain other information on what is in hands, an observation of nature, the data. In possession of the new application of the theory of dynamical systems, are enabled to us a new type of inference on the observed phenomenon, and its causes, until now hidden by the models. The technique of convergent-cross mapping, among attractors generated by phase-space reconstruction through the representation of the original phase-space in a Euclidean space formed by the original time series and its delays, enables us a more pragmatic inference of causality and more effective for systems that obey a nonlinear dynamics, the case for many ecological and biological series of interest. / 131659/2016-2.

Um dos principais cereais produzido e consumido no Brasil, o arroz se apresentou, principalmente a partir da década de 1990, como um dos produtos agrícolas mais importados pelo Brasil. A abertura econômica brasileira em 1990, o plano de estabilização monetária (Plano Real) em 1994, e a criação do Mercosul em 1995 possibilitaram uma maior importação de bens e mercadorias pelo Brasil. Não diferente, as importações de arroz do país saltaram a um patamar significativo ao longo da década de 1990, com o Uruguai e Argentina sendo os maiores ofertantes do produto no mercado brasileiro. Mesmo após sua desvalorização cambial em 1999, o Brasil manteve níveis consideráveis de importação do produto oriundo do Mercosul. De forma a compreender os fatores que contribuíram para um aumento da demanda do produto importado, o presente trabalho descreveu o cenário da orizicultura no Brasil, Argentina e Uruguai, desde 1989 a 2008, analisando suas cadeias produtivas, preços nos mercados domésticos, e participação no mercado internacional de arroz. Concomitante a isso, propôs-se um modelo econômico para analisar esta relação comercial no Mercosul, assumindo que as importações brasileiras de arroz são resultantes de um excesso de demanda doméstica pelo cereal. Utilizou-se um ferramental econométrico baseado em um Modelo Auto-regressivo Vetorial VAR estrutural, aplicando a análise às relações contemporâneas das quantidades de arroz importado do Mercosul pelo Brasil, do preço doméstico do arroz no Brasil, do preço de importação do arroz do Mercosul, da renda interna brasileira e da taxa de câmbio efetiva no Brasil. Os resultados são expressos a partir de uma matriz de relações contemporâneas, da decomposição da variância do erro de previsão e da função impulso-resposta das variáveis em relação a choques contemporâneos nas mesmas, através do processo de Bernanke. Os resultados alcançados mostraram uma forte relação entre o volume importado com o preço doméstico de arroz, além de uma relativa importância na taxa de câmbio brasileira na explicação do padrão de importação do arroz pelo país. Outro ponto importante é a significativa participação do preço de importação na explicação dos preços domésticos. A variável quantidade de importação de arroz se mostrou sensível em relação a um choque positivo de 1% simulado no preço doméstico, indicando um aumento imediato em aproximadamente 3% os volumes importados. Esta variável também se mostrou sensível a um choque no preço de importação, que apontou a uma redução pela metade (0,5%) no volume importado e, a um choque na taxa de câmbio, com redução de 2% no volume importado. Já os choques no preço doméstico e preço de importação do arroz indicaram uma relação de causalidade de um sobre o outro, sugerindo aumentos nos mesmos e indicando a existência de um mercado orizícola integrado entre o Brasil e o Mercosul. / The rice, one of the main cereals produced and consumed in Brazil, has been one of the most agricultural products imported by Brazil, mainly in the 1990s. The economic market opened in 1990, the monetary stabilization plan (Plano Real) in 1994 and the creation of Mercosur in 1995 allowed an increment in imports of goods and services in Brazil. Therefore, imports of rice in the country rose to a significant level throughout the 1990s. Uruguay and Argentina became the largest suppliers of the product for the Brazilian market. Even though, after the devaluation of the Brazilian currency (Real) in 1999, Brazil retained considerable levels of product imports from the Mercosur. In order to understand the factors that contributed to an increased demand of imported product, this thesis described the scenery of rice production in Brazil, Argentina and Uruguay, from 1989 to 2008. It analyses their production chains, prices in domestic markets, and participation in international rice´s market. Concomitant to this, it was proposed an economic model that examines the relationship in the Mercosur trade, assuming that Brazilian imports of rice are a result of excess domestic demand for this grain. Econometric model Vector Auto-regressive structural VAR was applied to analyze the contemporary relations of the quantities of rice imported by Brazil in Mercosur, the domestic price of rice in Brazil, the import price of rice in Mercosur, the Brazilian domestic income and the effective exchange rate in Brazil. Results are expressed from an array of contemporary relations, from decomposition variance of the prediction error and from the impulse-response function of the variables on contemporary shocks. The results showed a strong relationship between rice imports with the domestic rice price, and relative importance with Brazilian exchange rate in explaining the pattern of importing rice. Another point is the significant participation of the imports price in explaining domestic price. The variable quantity of rice imports was sensitive for a positive shock of 1% in the domestic price, indicating an immediate increment close to 3% in the quantity imported. This variable was also sensitive to a shock in the imported price, which showed a reduction by half (0.5%) in the quantity imported. A shock in the exchange rate reduces by 2% in the quantity imported. The shocks in the domestic and import price of rice indicated a causality relationship to one over the other, suggesting increases on both, indicating an existence of integrated rice market between Brazil and Mercosur.

O objetivo geral desta tese, composta por três artigos, é analisar as causas da criminalidade. As análises são feitas sob a ótica da Economia do Crime. Os dois primeiros artigos são independentes, mas complementares. Neles se busca lançar luz sobre as possíveis causas da queda do crime na cidade São Paulo. A ênfase recai na política de desarmamento dos cidadãos, no desempenho da Polícia e nas condições econômicas, em particular do mercado de trabalho. No primeiro deles, o objetivo é avaliar o efeito do Estatuto do Desarmamento sobre a criminalidade letal. Para isso foram utilizados dados de séries temporais da cidade de São Paulo na aplicação de uma metodologia de análise de intervenção. A hipótese de que a política de desarmamento causou redução na taxa de crimes letais não é rejeitada. Partindo dessa evidência, no segundo artigo o principal objetivo é investigar possíveis causas da significativa redução da criminalidade na cidade de São Paulo. Por meio de uma análise de cointegração evidenciaram-se relações de longo prazo entre crime, atividade econômica e desempenho da Polícia. Os resultados indicam que a taxa de crimes letais é positivamente relacionada ao desemprego, negativamente relacionada ao salário real e negativamente relacionada aos resultados das atividades de polícia, especificamente prisões e apreensão de armas de fogo. Ademais, não é rejeitada a hipótese de que o Estatuto do Desarmamento causou redução na taxa de crimes letais, reforçando a conclusão feita no primeiro artigo desta tese. No terceiro artigo, o foco das análises passa a ser os determinantes do risco de vitimização criminal. O objetivo é investigar os efeitos da riqueza dos indivíduos no risco de serem vítimas de crimes contra a propriedade, em particular crimes de furto/roubo a residência e furto/roubo a pessoa. Em termos específicos, o intuito é investigar se a relação entre risco de vitimização e riqueza pode ser descrita por uma parábola com concavidade voltada para baixo. São utilizados os dados de duas pesquisas domiciliares de vitimização realizadas na cidade de São Paulo na estimação de modelos probit. Os resultados das estimações indicaram que a riqueza dos indivíduos é um dos determinantes do risco de vitimização criminal a propriedade. Ademais, evidenciou-se que o risco de vitimização cresce com a riqueza, mas atinge um ponto de máximo, a partir do qual se reduz para níveis de riqueza mais elevados. / This three-article PhD thesis aims to investigate the causes of crime using an economic approach. The first two articles are independent, but complementary. In these articles, the objective is to shed light on possible causes of reductions in the crime rate in Sao Paulo city, focusing on the citizen disarmament policy, police performance, economic conditions and, particularly, the labor market. In the first article, the objective is assessing the effect of the Disarmament Statute on lethal crime rates. For this purpose, we used time-series data for Sao Paulo city in applying an intervention analysis methodology. The hypothesis that the disarmament policy led to a decline in the lethal crime rate is not rejected. Based on this evidence, the main objective of the second article is to investigate possible causes for the significant reduction observed in crime rates in Sao Paulo city. By applying a cointegration analysis, we observed long-run relationships between crime, economic activity and police performance. The results indicate that the lethal crime rate is positively related to unemployment and negatively related to real wages and to the results of law-enforcement activities, specifically arrests and seizure of firearms. Moreover, the hypothesis that the Disarmament Statute led to a reduction in the lethal crime rate is not rejected, reinforcing the conclusion arrived at in the first article of this thesis. In the third article, the focus of the analysis shifts to the determinants of criminal victimization. In this study, the objective is to investigate the effects of the wealth of individuals on the risk of becoming victims of property crimes, particularly crimes of theft/robbery of residence and theft/robbery of person. Specifically, its aim is to investigate whether the relationship between wealth and victimization risk can be described by a concave down parabola. Data from two household surveys on victimization held in Sao Paulo were used to estimate probit models. It became evident that the wealth of individuals is one of the determinants of victimization risk. And it was found that criminal victimization risk increases with wealth, but that it reaches a maximum point from which it decreases as wealth levels increase.