Previsão do índice bovespa por meio de redes neurais artificiais: uma análise comparada aos métodos tradicionais de séries de tempo

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 naïve 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

Modelagem e previsão da perda de injetividade em poços canhoneados

Waterflooding is a technique largely applied in the oil industry. The injected water displaces oil to the producer wells and avoid reservoir pressure decline. However, suspended particles in the injected water may cause plugging of pore throats causing formation damage (permeability reduction) and injectivity decline during waterflooding. When injectivity decline occurs it is necessary to increase the injection pressure in order to maintain water flow injection. Therefore, a reliable prediction of injectivity decline is essential in waterflooding projects. In this dissertation, a simulator based on the traditional porous medium filtration model (including deep bed filtration and external filter cake formation) was developed and applied to predict injectivity decline in perforated wells (this prediction was made from history data). Experimental modeling and injectivity decline in open-hole wells is also discussed. The injectivity of modeling showed good agreement with field data, which can be used to support plan stimulation injection wells

Um Modelo híbrido para previsão de curvas de produção de petróleo

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior

Previsão do módulo de elasticidade transversal de compósitos unidirecionais através de redes neurais mistas

The aim of this study is to create an artificial neural network (ANN) capable of modeling the transverse elasticity modulus (E2) of unidirectional composites. To that end, we used a dataset divided into two parts, one for training and the other for ANN testing. Three types of architectures from different networks were developed, one with only two inputs, one with three inputs and the third with mixed architecture combining an ANN with a model developed by Halpin-Tsai. After algorithm training, the results demonstrate that the use of ANNs is quite promising, given that when they were compared with those of the Halpín-Tsai mathematical model, higher correlation coefficient values and lower root mean square values were observed

Efeitos de tempestades magnéticas em sinais de gps, em Natal, Brasil

Satellites signals present disturbances (scintillations), due to presence of irregularities in the ionospheric plasma. In the present work, we dedicate to the study of the attenuation of these scintillations that is, an improvement in the signal, during the main magnetic storm phase during the period of October 2006 to February 2007. Using amplitude of scintillation 1.5GHz (L1) data of the net of satellites GPS, in the ionospheric station of Natal (5.84o S, 35.20o O, -20o dip) and geomagnetic indices, during the minimum solar cycle (referred to as cycle 23), demonstrating its anti-correlation between magnetic activity (Kp) and index of scintillation () (Bonelli2005 method, that works for solar maximum, Bonelli(2005)). These results show that these storms correspond to category I of Aarons (1991). The magnetic storms can generate irregularities when the electric feld of penetration eastward on the pre-reversal hour intensificating that and can too generate irregularities on midnight and sunlight period. The limitation of the method applied here is that it is not considering some storms that had also occurred during equinox and summer in the Brazilian region. For this reason, we will use additional data of the stations of São João de Cariri (Imager and Photometer) and of the station at Fortaleza (Digissonde data), as aid to analyze these storms. The storms that had been left out by the applied method, fit in the other two categories of Aarons, with one only exception (storm of January, 02). We show that in the day of the main phase of the magnetic storm, and with presence of bubbles (according to data of the Imager and Photometer), the speed of vertical drift ~E ~B (hF=t) is below 20m/s, that is the threshold found for Anderson al., (2004) (>20m/s as condition favorable to the formation of irregularities and increase in the scintillation index). This reduction of the speed is due to solar minimum

Estudo da previsão da carga de ruptura de estacas pré-moldadas de concreto

The piles are one of the most important types of solution adopted for the foundation of buildings. They are responsible for transmitting to the soil in deepe r and resistant layers loads from structures. The interaction of the foundation element with the soil is a very important variable, making indispensable your domain in order to determine the strength of the assembly and establish design criteria for each c ase of application of the pile. In this research analyzes were performed f rom experiments load tests for precast concrete piles and inve stigations of soil of type SPT, a study was performed for obtaining the ultimate load capacity of the foundation through methods extrapolation of load - settlement curve , semi - empirical and theoretic . After that, were realized comparisons between the different methods used for two types of soil a granular behavior and other cohesive. For obtaining soil paramet ers to be used i n the methods were established empirical correlations with the standard penetration number (NSPT). The charge - settlement curves of the piles are also analyzed. In the face of established comparisons was indicated the most reliable semiempirical method Déco urt - Quaresma as the most reliable for estimating the tensile strength for granular and cohesive soils. Meanwhile, among the methods studied extrapolation is recommended method of Van der Veen as the most appropriate for predicting the tensile strength.

Estudos de distúrbios ionosféricos propagantes de larga escala (LSTID’S) na região equatorial durante grandes tempestades magnéticas: estudo de casos

The Ionospheric Disturbances – TIDs – are irregularities on the ionospheric plasma propagating in speeds in the order of tens to a few hundreds of meters per second. This present study detected and characterized the TIDs of LSTIDs (Large Scale Travelling Ionospheric Disturbance) type at low latitudes during intense geomagnetic storms and its propagation over the Brazilian sector. This work also shows as being the first to report systematically propagation of gravity waves over Natal. For this purpose, we used ionospheric records obtained from type of digisonde CADI (Canadiam Advanced Digital Ionosonde) located in Natal and the type DSP (Digisonde Portable Souder) located in Cachoeira Paulista, Fortaleza and São Luis, whereupon we used a dataset of 12 years collected by INPE (National Institute of Space Research). In this study, both calm days, that preceded the storms, and the geomagnetically disturbed days were related during the years 2000 and 1012, which cover a period of maximum and minimum solar activity. And it is presented the variations that happened in the electron density from region F of the ionosphere over the Brazilian sector, especially near the Equator (Natal, Fortaleza and São Luis), caused by ionospheric disturbances in the equatorial region during intense geomagnetic storms, because, as we know of the literature in this area, this phenomenon contributes positively to the emergence of LSTIDs in the auroral region, which may move to the equatorial region where a few cases have been documented and studied systematically. From the observation of signatures if TIDs in ionogram records, a study of the morphology of these events was performed and compared with the main characteristics of the wave of this phenomenon during great magnetic storms, i.e., DST <(-200 nT) and KP > 6. Thus, we obtained the main characteristics of TIDs over our region, i.e., period, vertical wavelength, phase and propagation speed, as well as the delay of these disturbances compared to the beginning of the magnetic storms to the Brazilian Sector.

Estudo da variabilidade da deriva vertical do plasma ionosférico sobre a região de Natal/RN, Fortaleza/CE e Cachoeira Paulista/SP para alguns casos de tempestades magnéticas de abril a setembro de 2000

In this work are presented the results of research on the variation of vertical plasma drift from the F layer that commonly occurs during the period close to the going down of the sun (between 21UT - 22UT), and consisting of an increase in the F layer in time of day and a drop at night on the cities of Natal / RN (The 33.7º, 5.6º S), Fortaleza / CE (38.45º O, 3.9º S) and Cachoeira Paulista / SP (45.0ºO, 22.7º S ). Thus, data was used (ionograms) virtual height t ( h' ) as a function of the frequency F for the layer of the ionosphere. These data were collected through ionossonda type CADI (Canadian Advanced Digital Ionosonde) to the cities of Natal and Fortaleza, and the Digisonde DPS to the town of Cachoeira Paulista. They considered only the data available 2000 maximum solar year. With the virtual height ( h' ) data reduced in the frequencies of 6MHz and 5MHz, the next step was to calculate. The results showed that the variability of the vertical drift rate F ionospheric layer periods of magnetic storms is related both to the penetration of the electric fields as the disturbance dynamo electric fields.

Estimação e previsão no processo INARCH(2)

In the last decades the study of integer-valued time series has gained notoriety due to its broad applicability (modeling the number of car accidents in a given highway, or the number of people infected by a virus are two examples). One of the main interests of this area of study is to make forecasts, and for this reason it is very important to propose methods to make such forecasts, which consist of nonnegative integer values, due to the discrete nature of the data. In this work, we focus on the study and proposal of forecasts one, two and h steps ahead for integer-valued second-order autoregressive conditional heteroskedasticity processes [INARCH (2)], and in determining some theoretical properties of this model, such as the ordinary moments of its marginal distribution and the asymptotic distribution of its conditional least squares estimators. In addition, we study, via Monte Carlo simulation, the behavior of the estimators for the parameters of INARCH(2) processes obtained using three di erent methods (Yule- Walker, conditional least squares, and conditional maximum likelihood), in terms of mean squared error, mean absolute error and bias. We present some forecast proposals for INARCH(2) processes, which are compared again via Monte Carlo simulation. As an application of this proposed theory, we model a dataset related to the number of live male births of mothers living at Riachuelo city, in the state of Rio Grande do Norte, Brazil.

Estimação e previsão no processo INARCH(2)

In the last decades the study of integer-valued time series has gained notoriety due to its broad applicability (modeling the number of car accidents in a given highway, or the number of people infected by a virus are two examples). One of the main interests of this area of study is to make forecasts, and for this reason it is very important to propose methods to make such forecasts, which consist of nonnegative integer values, due to the discrete nature of the data. In this work, we focus on the study and proposal of forecasts one, two and h steps ahead for integer-valued second-order autoregressive conditional heteroskedasticity processes [INARCH (2)], and in determining some theoretical properties of this model, such as the ordinary moments of its marginal distribution and the asymptotic distribution of its conditional least squares estimators. In addition, we study, via Monte Carlo simulation, the behavior of the estimators for the parameters of INARCH(2) processes obtained using three di erent methods (Yule- Walker, conditional least squares, and conditional maximum likelihood), in terms of mean squared error, mean absolute error and bias. We present some forecast proposals for INARCH(2) processes, which are compared again via Monte Carlo simulation. As an application of this proposed theory, we model a dataset related to the number of live male births of mothers living at Riachuelo city, in the state of Rio Grande do Norte, Brazil.

Previsão do índice bovespa por meio de redes neurais artificiais: uma análise comparada aos métodos tradicionais de séries de tempo

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 naïve 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

Modelagem e previsão da perda de injetividade em poços canhoneados

Waterflooding is a technique largely applied in the oil industry. The injected water displaces oil to the producer wells and avoid reservoir pressure decline. However, suspended particles in the injected water may cause plugging of pore throats causing formation damage (permeability reduction) and injectivity decline during waterflooding. When injectivity decline occurs it is necessary to increase the injection pressure in order to maintain water flow injection. Therefore, a reliable prediction of injectivity decline is essential in waterflooding projects. In this dissertation, a simulator based on the traditional porous medium filtration model (including deep bed filtration and external filter cake formation) was developed and applied to predict injectivity decline in perforated wells (this prediction was made from history data). Experimental modeling and injectivity decline in open-hole wells is also discussed. The injectivity of modeling showed good agreement with field data, which can be used to support plan stimulation injection wells

Um Modelo híbrido para previsão de curvas de produção de petróleo

Coordenação de Aperfeiçoamento de Pessoal de Nível Superior