Orientador: Anna Diva Plasencia Lotufo / Resumo: O dióxido de carbono (CO2) é considerado um dos principais gases do efeito estufa adicional e contribui significativamente para as mudanças climáticas globais. Áreas agrícolas oferecem uma oportunidade para mitigar esse efeito, uma vez que, dependendo de seu uso e manejo, são capazes de armazenar grandes quantidades de carbono, retirando-as da atmosfera. A produção de CO2 no solo é resultado de processos biológicos, como a decomposição da matéria orgânica e respiração de raízes e organismos do solo, fenômeno chamado de emissão de CO2 do solo (FCO2). O objetivo deste trabalho foi utilizar as redes neurais artificiais para estudo e previsão de padrões espaço-temporais da emissão de CO2 do solo em áreas de cana-de-açúcar em sistema de cana crua, colheita mecanizada, quando grandes quantidades de palhas são depositadas sobre a superfície do solo. Valores de FCO2 foram coletados em áreas de cultivo comercial no Sudeste do Estado de São Paulo, registrados por meio do sistema LI-8100, em gradeados amostrais para determinação da variabilidade espaçotemporal de FCO2, e atributos físicos e químicos do solo. Foram utilizados dados referentes a estudos realizados nos anos de 2008, 2010 e 2012, no período após a operação de colheita mecânica da cultura. Uma rede neural Perceptron Multi-Camadas via algoritmo backpropagation foi aplicada para estimar a emissão de FCO2 do ano de 2012, utilizando os dados referentes aos anos de 2008 e 2010 para treinamento da rede neural. A rede neural inici... (Resumo completo, clicar acesso eletrônico abaixo) / Doutor

Este trabalho realiza uma análise exploratória do comportamento das venda de varejo na economia brasileira. O estudo parte da identificação de variáveis econômicas relevantes na determinação das vendas de varejo. A identificação apoia-se em proposições de outros estudos e testes de causalidade. O resultado da identificação é validado por meio da construção de um modelo de previsão de vendas no varejo. A relevância da causalidade das variáveis é verificada pela comparação do desempenho do modelo multivariado em relação ao desempenho de modelos univariados. Em seguida, o trabalho aprofunda a análise da relevância das variáveis por meio da mensuração da causalidade das taxas de juros básica da economia e ao consumidor e da mensuração do tempo médio de causalidade sobre as vendas de varejo. Por fim, as vendas de varejo são decompostas em vendas setoriais e avaliadas as mensurações de causalidade e tempo de causalidade das taxas de juros sobre os segmentos de varejo. / This work conducts an exploratory analysis on the behavior of retail sales in the brazilian economy. The analysis starts by identifying relevant economic variables in determining retail sales. The identification leans on other researches and causality tests. The identification results are validated by the contruction of a forecasting model for retail sales. The relevance of causal variables is assessed by comparing the forecasting performance of univariate models against the multivariate model. The work deepens the analysis of the causal variables by measuring the causality of the basic interest rate and the consumer interest rate over retail sales and by measuring the average time of the causality. The same analysis is extended to industry sectors in the retail sales.

Algae and cyanobacteria occur naturally in source waters and are known to cause extensive problems in the drinking water treatment industry. Cyanobacteria (especially Anabaena sp. and Microcystis sp.) are responsible for many water treatment problems in drinking water treatment works (DWTW) all over the world because of their ability to produce organic compounds like cyanotoxins (e.g. microcystin) and taste and odour compounds (e.g. geosmin) that can have an adverse effect on consumer health and consumer confidence in tap water. Therefore, the monitoring of cyanobacteria in source waters entering DWTW has become an essential part of drinking water treatment management. Managers of DWTW, rely heavily on results of physical, chemical and biological water quality analyses, for their management decisions. But results of water quality analyses can be delayed from 3 hours to a few days depending on a magnitude of factors such as: sampling, distance and accessibility to laboratory, laboratory sample turn-around times, specific methods used in analyses etc. Therefore the use of on-line (in situ) instruments that can supply real-time results by the click of a button has become very popular in the past few years. On-line instruments were developed for analyses like pH, conductivity, nitrate, chlorophyll-a and cyanobacteria concentrations. Although, this real-time (on-line) data has given drinking water treatment managers a better opportunity to make sound management decisions around drinking water treatment options based on the latest possible results, it may still be “too little, too late” once a sudden cyanobacterial bloom of especially Anabaena sp. or Microcystis sp. enters the plant. Therefore the benefit for drinking water treatment management, of changing the focus from real-time results to future predictions of water quality has become apparent. The aims of this study were 1) to review the environmental variables associated with cyanobacterial blooms in the Vaal Dam, as to get background on the input variables that can be used in cyanobacterial-related forecasting models; 2) to apply rule-based Hybrid Evolutionary Algorithms (HEAs) to develop models using a) all applicable laboratory-generated data and b) on-line measureable data only, as input variables in prediction models for harmful algal blooms in the Vaal Dam; 3) to test these models with data that was not used to develop the models (so-called “unseen data”), including on-line (in situ) generated data; and 4) to incorporate selected models into two cyanobacterial incident management protocols which link to the Water Safety Plan (WSP) of a large DWTW (case study : Rand Water). During the current study physical, chemical and biological water quality data from 2000 to 2009, measured in the Vaal Dam and the 20km long canal supplying the Zuikerbosch DWTW of Rand Water, has been used to develop models for the prediction of Anabaena sp., Microcystis sp., the cyanotoxin microcystin and the taste and odour compound geosmin for different prediction or forecasting times in the source water. For the development and first stage of testing the models, 75% of the dataset was used to train the models and the remaining 25% of the dataset was used to test the models. Boot-strapping was used to determine which 75% of the dataset was to be used as the training dataset and which 25% as the testing dataset. Models were also tested with 2 to 3 years of so called “unseen data” (Vaal Dam 2010 – 2012) i.e. data not used at any stage during the model development. Fifty different models were developed for each set of “x input variables = 1 output variable” chosen beforehand. From the 50 models, the best model between the measured data and the predicted data was chosen. Sensitivity analyses were also performed on all input variables to determine the variables that have the largest impact on the result of the output. This study have shown that hybrid evolutionary algorithms can successfully be used to develop relatively accurate forecasting models, which can predict cyanobacterial cell concentrations (particularly Anabaena sp. and Microcystis sp.), as well as the cyanotoxin microcystin concentration in the Vaal Dam, for up to 21 days in advance (depending on the output variable and the model applied). The forecasting models that performed the best were those forecasting 7 days in advance (R2 = 0.86, 0.91 and 0.75 for Anabaena[7], Microcystis[7] and microcystin[7] respectively). Although no optimisation strategies were performed, the models developed during this study were generally more accurate than most models developed by other authors utilising the same concepts and even models optimised by hill climbing and/or differential evolution. It is speculated that including “initial cyanobacteria inoculum” as input variable (which is unique to this study), is most probably the reason for the better performing models. The results show that models developed from on-line (in situ) measureable data only, are almost as good as the models developed by using all possible input variables. The reason is most probably because “initial cyanobacteria inoculum” – the variable towards which the output result showed the greatest sensitivity – is included in these models. Generally models predicting Microcystis sp. in the Vaal Dam were more accurate than models predicting Anabaena sp. concentrations and models with a shorter prediction time (e.g. 7 days in advance) were statistically more accurate than models with longer prediction times (e.g. 14 or 21 days in advance). The multi-barrier approach in risk reduction, as promoted by the concept of water safety plans under the banner of the Blue Drop Certification Program, lends itself to the application of future predictions of water quality variables. In this study, prediction models of Anabaena sp., Microcystis sp. and microcystin concentrations 7 days in advance from the Vaal Dam, as well as geosmin concentration 7 days in advance from the canal were incorporated into the proposed incident management protocols. This was managed by adding an additional “Prediction Monitoring Level” to Rand Waters’ microcystin and taste and odour incident management protocols, to also include future predictions of cyanobacteria (Anabaena sp. and Microcystis sp.), microcystin and geosmin. The novelty of this study was the incorporation of future predictions into the water safety plan of a DWTW which has never been done before. This adds another barrier in the potential exposure of drinking water consumers to harmful and aesthetically unacceptable organic compounds produced by cyanobacteria. / PhD (Botany), North-West University, Potchefstroom Campus, 2015

Algae and cyanobacteria occur naturally in source waters and are known to cause extensive problems in the drinking water treatment industry. Cyanobacteria (especially Anabaena sp. and Microcystis sp.) are responsible for many water treatment problems in drinking water treatment works (DWTW) all over the world because of their ability to produce organic compounds like cyanotoxins (e.g. microcystin) and taste and odour compounds (e.g. geosmin) that can have an adverse effect on consumer health and consumer confidence in tap water. Therefore, the monitoring of cyanobacteria in source waters entering DWTW has become an essential part of drinking water treatment management. Managers of DWTW, rely heavily on results of physical, chemical and biological water quality analyses, for their management decisions. But results of water quality analyses can be delayed from 3 hours to a few days depending on a magnitude of factors such as: sampling, distance and accessibility to laboratory, laboratory sample turn-around times, specific methods used in analyses etc. Therefore the use of on-line (in situ) instruments that can supply real-time results by the click of a button has become very popular in the past few years. On-line instruments were developed for analyses like pH, conductivity, nitrate, chlorophyll-a and cyanobacteria concentrations. Although, this real-time (on-line) data has given drinking water treatment managers a better opportunity to make sound management decisions around drinking water treatment options based on the latest possible results, it may still be “too little, too late” once a sudden cyanobacterial bloom of especially Anabaena sp. or Microcystis sp. enters the plant. Therefore the benefit for drinking water treatment management, of changing the focus from real-time results to future predictions of water quality has become apparent. The aims of this study were 1) to review the environmental variables associated with cyanobacterial blooms in the Vaal Dam, as to get background on the input variables that can be used in cyanobacterial-related forecasting models; 2) to apply rule-based Hybrid Evolutionary Algorithms (HEAs) to develop models using a) all applicable laboratory-generated data and b) on-line measureable data only, as input variables in prediction models for harmful algal blooms in the Vaal Dam; 3) to test these models with data that was not used to develop the models (so-called “unseen data”), including on-line (in situ) generated data; and 4) to incorporate selected models into two cyanobacterial incident management protocols which link to the Water Safety Plan (WSP) of a large DWTW (case study : Rand Water). During the current study physical, chemical and biological water quality data from 2000 to 2009, measured in the Vaal Dam and the 20km long canal supplying the Zuikerbosch DWTW of Rand Water, has been used to develop models for the prediction of Anabaena sp., Microcystis sp., the cyanotoxin microcystin and the taste and odour compound geosmin for different prediction or forecasting times in the source water. For the development and first stage of testing the models, 75% of the dataset was used to train the models and the remaining 25% of the dataset was used to test the models. Boot-strapping was used to determine which 75% of the dataset was to be used as the training dataset and which 25% as the testing dataset. Models were also tested with 2 to 3 years of so called “unseen data” (Vaal Dam 2010 – 2012) i.e. data not used at any stage during the model development. Fifty different models were developed for each set of “x input variables = 1 output variable” chosen beforehand. From the 50 models, the best model between the measured data and the predicted data was chosen. Sensitivity analyses were also performed on all input variables to determine the variables that have the largest impact on the result of the output. This study have shown that hybrid evolutionary algorithms can successfully be used to develop relatively accurate forecasting models, which can predict cyanobacterial cell concentrations (particularly Anabaena sp. and Microcystis sp.), as well as the cyanotoxin microcystin concentration in the Vaal Dam, for up to 21 days in advance (depending on the output variable and the model applied). The forecasting models that performed the best were those forecasting 7 days in advance (R2 = 0.86, 0.91 and 0.75 for Anabaena[7], Microcystis[7] and microcystin[7] respectively). Although no optimisation strategies were performed, the models developed during this study were generally more accurate than most models developed by other authors utilising the same concepts and even models optimised by hill climbing and/or differential evolution. It is speculated that including “initial cyanobacteria inoculum” as input variable (which is unique to this study), is most probably the reason for the better performing models. The results show that models developed from on-line (in situ) measureable data only, are almost as good as the models developed by using all possible input variables. The reason is most probably because “initial cyanobacteria inoculum” – the variable towards which the output result showed the greatest sensitivity – is included in these models. Generally models predicting Microcystis sp. in the Vaal Dam were more accurate than models predicting Anabaena sp. concentrations and models with a shorter prediction time (e.g. 7 days in advance) were statistically more accurate than models with longer prediction times (e.g. 14 or 21 days in advance). The multi-barrier approach in risk reduction, as promoted by the concept of water safety plans under the banner of the Blue Drop Certification Program, lends itself to the application of future predictions of water quality variables. In this study, prediction models of Anabaena sp., Microcystis sp. and microcystin concentrations 7 days in advance from the Vaal Dam, as well as geosmin concentration 7 days in advance from the canal were incorporated into the proposed incident management protocols. This was managed by adding an additional “Prediction Monitoring Level” to Rand Waters’ microcystin and taste and odour incident management protocols, to also include future predictions of cyanobacteria (Anabaena sp. and Microcystis sp.), microcystin and geosmin. The novelty of this study was the incorporation of future predictions into the water safety plan of a DWTW which has never been done before. This adds another barrier in the potential exposure of drinking water consumers to harmful and aesthetically unacceptable organic compounds produced by cyanobacteria. / PhD (Botany), North-West University, Potchefstroom Campus, 2015

The presented study describes commercial real estate markets with focus on office sector. We identify the capitalization rate (investment yield) as one of the fundamental elements in the commercial property valuation. Based on historical office investment yield observations and various econometric models we predict the office capitalization rate development in the Czech Republic. We use data of the United Kingdom, Ireland and Sweden to identify common yield trend especially with respect to their real estate crises in 1990s that embody features similar to the real estate crisis in 2008-2010. As explanatory variables for the econometric models (ARIMA, OLS, VAR) we use financial and macroeconomic variables. We use the OLS models to identify the optimal set of explanatory variables, to be applied in VAR models. On dataset of the comparable countries we compare the goodness of fit of the VAR and ARIMA models. The best variants are then used for the prediction of the Czech office yield. Lastly, we improve our results by implementing exogenous forecasts of macroeconomic variables used in the models. Majority of our predictions forecast a slow decrease of the capitalization rate in next two years (2010-2012) in the magnitude of 0.25% - 1% (to 6.25%-6%).

Made available in DSpace on 2014-12-17T13:53:32Z (GMT). No. of bitstreams: 1 RenataLRS_DISSERT.pdf: 1647146 bytes, checksum: 4d5eb3f745488991eeacb24559330562 (MD5) Previous issue date: 2011-12-20 / Forecast is the basis for making strategic, tactical and operational business decisions. In financial economics, several techniques have been used to predict the behavior of assets over the past decades.Thus, there are several methods to assist in the task of time series forecasting, however, conventional modeling techniques such as statistical models and those based on theoretical mathematical models have produced unsatisfactory predictions, increasing the number of studies in more advanced methods of prediction. Among these, the Artificial Neural Networks (ANN) are a relatively new and promising method for predicting business that shows a technique that has caused much interest in the financial environment and has been used successfully in a wide variety of financial modeling systems applications, in many cases proving its superiority over the statistical models ARIMA-GARCH. In this context, this study aimed to examine whether the ANNs are a more appropriate method for predicting the behavior of Indices in Capital Markets than the traditional methods of time series analysis. For this purpose we developed an quantitative study, from financial economic indices, and developed two models of RNA-type feedfoward supervised learning, whose structures consisted of 20 data in the input layer, 90 neurons in one hidden layer and one given as the output layer (Ibovespa). These models used backpropagation, an input activation function based on the tangent sigmoid and a linear output function. Since the aim of analyzing the adherence of the Method of Artificial Neural Networks to carry out predictions of the Ibovespa, we chose to perform this analysis by comparing results between this and Time Series Predictive Model GARCH, developing a GARCH model (1.1).Once applied both methods (ANN and GARCH) we conducted the results' analysis by comparing the results of the forecast with the historical data and by studying the forecast errors by the MSE, RMSE, MAE, Standard Deviation, the Theil's U and forecasting encompassing tests. It was found that the models developed by means of ANNs had lower MSE, RMSE and MAE than the GARCH (1,1) model and Theil U test indicated that the three models have smaller errors than those of a na?ve forecast. Although the ANN based on returns have lower precision indicator values than those of ANN based on prices, the forecast encompassing test rejected the hypothesis that this model is better than that, indicating that the ANN models have a similar level of accuracy . It was concluded that for the data series studied the ANN models show a more appropriate Ibovespa forecasting than the traditional models of time series, represented by the GARCH model / Nas organiza??es, a previs?o constitui a base para a tomada de decis?es estrat?gicas, t?ticas e operacionais. Na economia financeira, diversas t?cnicas t?m sido usadas a fim de prever o comportamento de ativos no decorrer das ?ltimas d?cadas. Assim, existem diversos m?todos para auxiliar na tarefa de previs?o de s?ries temporais, entretanto, t?cnicas de modelagem convencionais como modelos estat?sticos e aqueles baseados em modelos matem?ticos te?ricos t?m produzido previs?es insatisfat?rias, aumentando o n?mero de estudos em m?todos mais avan?ados de previs?o. Dentre estes, as Redes Neurais Artificiais (RNA) s?o um m?todo relativamente recente e promissor para a previs?o em neg?cios que se revela uma das t?cnicas que tem causado muito interesse no ambiente financeiro e tem sido utilizado com sucesso em uma ampla variedade de aplica??es de sistemas de modelagem financeiro, provado em muitos casos sua superioridade sobre os modelos estat?sticos ARIMA-GARCH (OLIVEIRA,2007). Nesse contexto, o presente trabalho teve por objetivo analisar se as RNAs s?o um m?todo mais adequado para a previs?o do comportamento de ?ndices em Mercados de Capital do que m?todos tradicionais de an?lise de s?ries temporais. Para tanto, foi desenvolvido um estudo quantitativo que, a partir de ?ndices econ?mico financeiros, elaborou dois modelos de RNA do tipo feedfoward de aprendizado supervisionado, cujas estruturas consistiram em 20 dados na camada de entrada, 90 neur?nios em uma camada oculta e um dado como camada de sa?da (?ndice Ibovespa). Estes modelos utilizaram BackPropagation, fun??o de ativa??o de entrada baseada na tangente Sigmoid e uma fun??o de sa?da linear. Visto o intuito de analisar a ader?ncia do M?todo de Redes Neurais Artificiais ? realiza??o de previs?es do Ibovespa, optou-se por realizar tal an?lise por meio da compara??o de resultados entre este e o M?todo de previs?o em s?ries temporais GARCH, desenvolvendo-se um modelo GARCH (1,1). Uma vez aplicadas ambas as metodologias (RNA e GARCH) e desenvolvidos os modelos, realizou-se a an?lise dos resultados obtidos comparando-se os resultados das previs?es com os dados hist?ricos e estudando-se os erros de previs?o por meio do MSE, RMSE, MAE, Desvio Padr?o, U de Theil e teste abrangente da previs?es. Verificou-se que os modelos desenvolvidos por meio de RNAs apresentaram menores MSE, RMSE e MAE que o modelo de controle e o teste U de Theil indicou que os tr?s modelos estudados apresentam erros menores que os de uma previs?o ing?nua. Embora a RNA baseada em retornos tenha apresentado valores dos indicadores de precis?o inferiores aos da RNA baseada em pre?os, o teste abrangente de regress?es rejeitou a hip?tese de que este modelo seja superior que aquele, indicando que os modelos de RNA apresentam um n?vel semelhante de precis?o. Concluiu-se que, para a s?rie de dados estudada neste trabalho, as Redes Neurais artificiais se mostram um modelo mais adequado de previs?o do que os modelos tradicionais de s?ries temporais, representado neste pelo m?todo GARCH

Universidade Federal de Uberlândia / The performance ratios have an important role inside and outside the companies. Internally, they are responsible for operationalise and measure each specific objective tracked by managers. Externally, they serve as flags of the situation of the company to stakeholders, influencing the market value of the organization. Therefore, an analysis through the use of ratios, using accounting, technical and market data from the company, can contribute to maximize their value and the shareholders earnings. But such analysis are not consensus on the financial literature. There are ways to be travelled, so far. One of these ways is a combination of business performance pointed by ratios with the future market value of these companies, trying to verify the validity of these ratios to provide valuations and / or devaluation of shares in a two years time period. This paper s objective is to verify the use of performance ratios functionality - fundamentalists, capital structure, liquidity, activity, profitability, market and technical ratios - for companies from different countries and industries considering the last three years to predict the market value for the next two years. It uses a quantitative research, which demands statistical techniques and multivariate analysis (multiple linear regression and binary logistic regression). For the goals it is a descriptive research. As technical procedures used, it can be classified into two categories: documentary and expost-fact research. The results show that, generally, the performance ratios - fundamentalists, capital structure, liquidity, activity, profitability, market and technical ratios - of the last three years can predict satisfactorily the market value of the next two years. The exception is made for Finance and Insurance industry. The ratios that are more related to the future market value of the corporations were the: Average from 2002 to 2004 of Enterprise Value, Average from 2002 to 2004 of EBIT + Net financial expenses, Range from 2002 to 2004 of the income / Stock Price, Average from 2002 to 2004 of EBITDA, Average from 2002 to 2004 of Capital Employed, Average from 2002 to 2004 of Floating Capital, Range from 2002 to 2004 of Return over Assets, Range from 2002 to 2004 of Gross Margin, Range from 2002 to 2004 of Sales to Stock, Range from 2002 to 2004 of Investment / equity. The main groups that are related with the future market value are: profitability, capital structure and market ratios. All the logistic regression models (for the set of companies and by industries) can discern over 60% of evaluation and devaluation cases, which is considered to be satisfactory. These results are limited to an specific historical data performance ratios for the past three years (2002 to 2004) and the market value of the subsequent years (2005 to 2006) and may not be generalized to other periods or set of periods. For future papers it is intended, primarily, to explain the problem presented by logistic regression models, unable to classify the majority devaluation cases on the next two years; verify the applicability of these results to other years and set of periods and separate the analysis by countries to verify the efficiency of each market. / Os indicadores de desempenho têm um papel importante dentro e fora das empresas. Internamente, são responsáveis por operacionalizar e medir cada objetivo específico traçado pelos gestores. Externamente, servem como sinalizadores da situação da empresa aos stakeholders, influenciando o valor de mercado da organização. Neste sentido, uma análise através da utilização de indicadores de dados contábeis, técnicos e de mercado da empresa, pode contribuir tanto para a empresa maximizar seu valor quanto para o acionista maximizar seus ganhos. Porém, tal tautologia não é consenso na literatura financeira. Isto porque ainda existem caminhos a serem percorridos. Um destes caminhos é a associação do desempenho empresarial apontado pelos indicadores com uma situação futura de valor de mercado destas empresas, pretendendo-se verificar a validade destes indicadores para se prever valorizações e/ou desvalorizações de ações em um horizonte de dois anos. Neste sentido, este trabalho visa verificar a funcionalidade da utilização de indicadores de desempenho - fundamentalistas, de estrutura de capital, de liquidez, de atividade, de rentabilidade, de mercado e técnicos para empresas de diferentes países e setores dos três últimos anos para prever o valor de mercado para os próximos dois anos. Para tanto foi utilizada, quanto à forma de abordagem, uma pesquisa quantitativa, que utiliza-se de técnicas estatísticas e de análise multivariada (regressão linear múltipla e regressão logística binária). Quanto aos objetivos, trata-se de uma pesquisa descritiva. Quanto aos procedimentos técnicos adotados, esta pesquisa pode ser classificados em duas categorias: pesquisa documental e expost-facto. Os resultados obtidos de modo geral apontam que os indicadores de desempenho - fundamentalistas, de estrutura de capital, de liquidez, de atividade, de rentabilidade, de mercado e técnicos dos três últimos anos conseguem prever satisfatoriamente o valor de mercado dos próximos dois anos. A exceção é feita ao setor de Finanças e Seguros. Os indicadores que mais se relacionaram com o valor de mercado futuro das Sociedades Anônimas foram: Média de 2002 a 2004 do Enterprise Value, Média de 2002 a 2004 do LAIR + Despesas Financeiras Líquidas, Variação de 2002 a 2004 do Lucro / Preço da Ação, Média de 2002 a 2004 do EBITDA, Média de 2002 a 2004 do Capital Employed, Média de 2002 a 2004 do Capital de Giro, Variação de 2002 a 2004 da Rentabilidade do Ativo, Variação de 2002 a 2004 da Margem Bruta, Variação de 2002 a 2004 das Vendas por Ação, Variação de 2002 a 2004 dos Investimentos / Patrimônio Líquido. Os principais grupos que se relacionam com o valor de mercado futuro são os de Rentabilidade, Estrutura de capital e Mercado. Na classificação global dos modelos de regressão logística todos os modelos (para o conjunto de Sociedades Anônimas e por setores) conseguiram distinguir acima de 60% dos casos de valorização e desvalorização, o que pode ser considerado satisfatório. Estes resultados são limitados aos dados históricos dos indicadores de desempenho dos últimos três anos (2002 a 2004) e ao valor de mercado dos anos subseqüentes (2005 a 2006), não podendo ser generalizados para outros períodos ou agregação de períodos. Para trabalhos futuros pretende-se, principalmente, explicar o problema apresentado pelos modelos de regressão logística de não conseguirem classificar a maioria dos casos de empresas que desvalorizarão nos próximos dois anos; verificar a aplicabilidade destes resultados aos demais anos e agrupamento de períodos e separar a análise por países para verificar a eficiência de cada mercado. / Mestre em Administração

Submitted by (edna.saturno@ufrpe.br) on 2017-03-30T14:18:20Z No. of bitstreams: 1 Ricardo Tavares Antunes de Oliveira.pdf: 797561 bytes, checksum: ec9749d5a2c4b94162d3aa69821c19b0 (MD5) / Made available in DSpace on 2017-03-30T14:18:20Z (GMT). No. of bitstreams: 1 Ricardo Tavares Antunes de Oliveira.pdf: 797561 bytes, checksum: ec9749d5a2c4b94162d3aa69821c19b0 (MD5) Previous issue date: 2015-02-26 / This dissertation disusses the problem of combining models for time series forecasting. In this context, the main characteristics and basic properties of combined models are presented. Some of the main methods of time series forecasting present in the literature are described. Computational experiments compare diverse copulas-based combined models. First of all, an algorithm is presented to combine predictions of the predictive models via Gumbel-Hougaard copula. In the second case, it is proposed a combined estimator constructed via non parametric Cacoullos multivariate functions. In the third and nal case of study, the main results of this dissertation are presented, in which an experiment that compares combined estimators constructed taking into account thousands of time series and numerous forecasting models were simulated. Thus, computational experiments show that the combined estimator constructed via copula obtained better results compared with the individual models and the linear combination method. / Sob o prisma da Estatí stica-Computacional, esta disserta ção trata do problema de combinação não linear de modelos de previsão de s éries temporais. Neste contexto, suas principais caracterí sticas e propriedades b ásicas são apresentadas. Alguns dos principais m étodos de previsão de s éries temporais presentes na literatura são descritos. As propostas apresentadas nesta disserta ção são mostradas por meio de três casos de estudo utilizando um formalismo matem ático conhecido como c ópulas pela sua capacidade de poder medir dependência e combinar modelos de predi ção sem perca de informa ção. A combina ção dos modelos de previsão ocorre por meio do formalismo matem ático de c ópulas que apresenta resultados convincentes e motivadores atrav és de três casos de estudo presentes nesta disserta ção. No primeiro é apresentado um pequeno caso para combinar as previsões dos modelos de previsão via c ópula de Gumbel-Hougaard. No segundo caso de estudo e proposto um estimador combinado constru ído atrav és da fun ção multivariada não-param étrica de Cacoullos para um caso simples de combina ção. No terceiro e último caso de estudo são apresentados os principais resultados desta disserta ção, em que, é realizado um experimento que compara estimadores combinados construí dos levando em considera ção v árias s éries temporais e in úmeros modelos de previsão, tal que, são simuladas diversas situa ções. Nesse sentido, experimentos computacionais realizados demostram que o estimador combinado construí do via c ópula obteve melhores resultados quando comparado com os modelos individuais e o m étodo de combina ção linear.

The aim of the thesis is through the selected analysis analyze the functioning and financing of health centre.

Traditional forecasting models have been widely used for decision-making in production, finance and energy. Such is the case of the ARIMA models, developed in the 1970s by George Box and Gwilym Jenkins [1], which incorporate characteristics of the past models of the same series, according to their autocorrelation. This work compares advanced statistical methods for determining the demand for electricity in Colombia, including the SARIMA, econometric and Bayesian methods.