The combination of neural estimates in prediction and decision problems

In this dissertation, different ways of combining neural predictive models or neural-based forecasts are discussed. The proposed approaches consider mostly Gaussian radial basis function networks, which can be efficiently identified and estimated through recursive/adaptive methods. Two different ways of combining are explored to get a final estimate – model mixing and model synthesis –, with the aim of obtaining improvements both in terms of efficiency and effectiveness. In the context of model mixing, the usual framework for linearly combining estimates from different models is extended, to deal with the case where the forecast errors from those models are correlated. In the context of model synthesis, and to address the problems raised by heavily nonstationary time series, we propose hybrid dynamic models for more advanced time series forecasting, composed of a dynamic trend regressive model (or, even, a dynamic harmonic regressive model), and a Gaussian radial basis function network. Additionally, using the model mixing procedure, two approaches for decision-making from forecasting models are discussed and compared: either inferring decisions from combined predictive estimates, or combining prescriptive solutions derived from different forecasting models. Finally, the application of some of the models and methods proposed previously is illustrated with two case studies, based on time series from finance and from tourism.

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

Dynamical aspects of atmospheric data assimilation in the tropics

A faithful depiction of the tropical atmosphere requires three-dimensional sets of observations. Despite the increasing amount of observations presently available, these will hardly ever encompass the entire atmosphere and, in addition, observations have errors. Additional (background) information will always be required to complete the picture. Valuable added information comes from the physical laws governing the flow, usually mediated via a numerical weather prediction (NWP) model. These models are, however, never going to be error-free, why a reliable estimate of their errors poses a real challenge since the whole truth will never be within our grasp. The present thesis addresses the question of improving the analysis procedures for NWP in the tropics. Improvements are sought by addressing the following issues: - the efficiency of the internal model adjustment, - the potential of the reliable background-error information, as compared to observations, - the impact of a new, space-borne line-of-sight wind measurements, and - the usefulness of multivariate relationships for data assimilation in the tropics. Most NWP assimilation schemes are effectively univariate near the equator. In this thesis, a multivariate formulation of the variational data assimilation in the tropics has been developed. The proposed background-error model supports the mass-wind coupling based on convectively-coupled equatorial waves. The resulting assimilation model produces balanced analysis increments and hereby increases the efficiency of all types of observations. Idealized adjustment and multivariate analysis experiments highlight the importance of direct wind measurements in the tropics. In particular, the presented results confirm the superiority of wind observations compared to mass data, in spite of the exact multivariate relationships available from the background information. The internal model adjustment is also more efficient for wind observations than for mass data. In accordance with these findings, new satellite wind observations are expected to contribute towards the improvement of NWP and climate modeling in the tropics. Although incomplete, the new wind-field information has the potential to reduce uncertainties in the tropical dynamical fields, if used together with the existing satellite mass-field measurements. The results obtained by applying the new background-error representation to the tropical short-range forecast errors of a state-of-art NWP model suggest that achieving useful tropical multivariate relationships may be feasible within an operational NWP environment.

A dynamical system approach to data assimilation in chaotic models

The Assimilation in the Unstable Subspace (AUS) was introduced by Trevisan and Uboldi in 2004, and developed by Trevisan, Uboldi and Carrassi, to minimize the analysis and forecast errors by exploiting the flow-dependent instabilities of the forecast-analysis cycle system, which may be thought of as a system forced by observations. In the AUS scheme the assimilation is obtained by confining the analysis increment in the unstable subspace of the forecast-analysis cycle system so that it will have the same structure of the dominant instabilities of the system. The unstable subspace is estimated by Breeding on the Data Assimilation System (BDAS). AUS- BDAS has already been tested in realistic models and observational configurations, including a Quasi-Geostrophicmodel and a high dimensional, primitive equation ocean model; the experiments include both fixed and“adaptive”observations. In these contexts, the AUS-BDAS approach greatly reduces the analysis error, with reasonable computational costs for data assimilation with respect, for example, to a prohibitive full Extended Kalman Filter. This is a follow-up study in which we revisit the AUS-BDAS approach in the more basic, highly nonlinear Lorenz 1963 convective model. We run observation system simulation experiments in a perfect model setting, and with two types of model error as well: random and systematic. In the different configurations examined, and in a perfect model setting, AUS once again shows better efficiency than other advanced data assimilation schemes. In the present study, we develop an iterative scheme that leads to a significant improvement of the overall assimilation performance with respect also to standard AUS. In particular, it boosts the efficiency of regime’s changes tracking, with a low computational cost. Other data assimilation schemes need estimates of ad hoc parameters, which have to be tuned for the specific model at hand. In Numerical Weather Prediction models, tuning of parameters — and in particular an estimate of the model error covariance matrix — may turn out to be quite difficult. Our proposed approach, instead, may be easier to implement in operational models.

A comparison of techniques for forecasting utilization of Veterans Affairs hospitals

This study demonstrated that accurate, short-term forecasts of Veterans Affairs (VA) hospital utilization can be made using the Patient Treatment File (PTF), the inpatient discharge database of the VA. Accurate, short-term forecasts of two years or less can reduce required inventory levels, improve allocation of resources, and are essential for better financial management. These are all necessary achievements in an era of cost-containment.^ Six years of non-psychiatric discharge records were extracted from the PTF and used to calculate four indicators of VA hospital utilization: average length of stay, discharge rate, multi-stay rate (a measure of readmissions) and days of care provided. National and regional levels of these indicators were described and compared for fiscal year 1984 (FY84) to FY89 inclusive.^ Using the observed levels of utilization for the 48 months between FY84 and FY87, five techniques were used to forecast monthly levels of utilization for FY88 and FY89. Forecasts were compared to the observed levels of utilization for these years. Monthly forecasts were also produced for FY90 and FY91.^ Forecasts for days of care provided were not produced. Current inpatients with very long lengths of stay contribute a substantial amount of this indicator and it cannot be accurately calculated.^ During the six year period between FY84 and FY89, average length of stay declined substantially, nationally and regionally. The discharge rate was relatively stable, while the multi-stay rate increased slightly during this period. FY90 and FY91 forecasts show a continued decline in the average length of stay, while the discharge rate is forecast to decline slightly and the multi-stay rate is forecast to increase very slightly.^ Over a 24 month ahead period, all three indicators were forecast within a 10 percent average monthly error. The 12-month ahead forecast errors were slightly lower. Average length of stay was less easily forecast, while the multi-stay rate was the easiest indicator to forecast.^ No single technique performed significantly better as determined by the Mean Absolute Percent Error, a standard measure of error. However, Autoregressive Integrated Moving Average (ARIMA) models performed well overall and are recommended for short-term forecasting of VA hospital utilization. ^

Constraining storm-scale forecasts of deep convective initiation with surface weather observations

Thesis (Ph.D.)--University of Washington, 2016-06

Eliciting and Aggregating Forecasts When Information is Shared

Using the wisdom of crowds---combining many individual forecasts to obtain an aggregate estimate---can be an effective technique for improving forecast accuracy. When individual forecasts are drawn from independent and identical information sources, a simple average provides the optimal crowd forecast. However, correlated forecast errors greatly limit the ability of the wisdom of crowds to recover the truth. In practice, this dependence often emerges because information is shared: forecasters may to a large extent draw on the same data when formulating their responses.

To address this problem, I propose an elicitation procedure in which each respondent is asked to provide both their own best forecast and a guess of the average forecast that will be given by all other respondents. I study optimal responses in a stylized information setting and develop an aggregation method, called pivoting, which separates individual forecasts into shared and private information and then recombines these results in the optimal manner. I develop a tailored pivoting procedure for each of three information models, and introduce a simple and robust variant that outperforms the simple average across a variety of settings.

In three experiments, I investigate the method and the accuracy of the crowd forecasts. In the first study, I vary the shared and private information in a controlled environment, while the latter two studies examine forecasts in real-world contexts. Overall, the data suggest that a simple minimal pivoting procedure provides an effective aggregation technique that can significantly outperform the crowd average.

Effect of wind on operating cost based cruise speed reduction for delay absorption

En route speed reduction can be used for air traffic flow management (ATFM), e.g., delaying aircraft while airborne or realizing metering at an arrival fix. In previous publications, the authors identified the flight conditions that maximize the airborne delay without incurring extra fuel consumption with respect to the nominal (not delayed) flight. In this paper, the effect of wind on this strategy is studied, and the sensitivity to wind forecast errors is also assessed. A case study done in Chicago O’Hare airport (ORD) is presented, showing that wind has a significant effect on the airborne delay that can be realized and that, in some cases, even tailwinds might lead to an increase in the maximum amount of airborne delay. The values of airborne delay are representative enough to suggest that this speed reduction technique might be useful in a real operational scenario. Moreover, the speed reduction strategy is more robust than nominal operations against fuel consumption in the presence of wind forecast uncertainties.

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

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

INFLACIÓN E INCERTIDUMBRE INFLACIONARIA: EVIDENCIA PARA COSTA RICA

En este documento se estima una medida de la incertidumbre inflacionaria. Un modelo de inflación señala incertidumbre cuando los errores de pronóstico son heteroscedásticos. Por medio de la especificación de una ecuación GARCH (Generalized Autoregressive Conditional Heteroscedasticity), para la varianza del término de error de un modelo de inflación, es posible estimar una proxy de incertidumbre inflacionaria. La estimación simultánea del modelo de inflación y de la ecuación GARCH, produce un nuevo modelo de inflación en el cual los errores de pronóstico son homocedásticos. Existe consenso en la literatura económica en que hay una correlación positiva entre incertidumbre inflacionaria y la magnitud de la tasa de inflación, lo cual, como lo señaló Friedman (1977), representa uno de los costos asociados con la persistencia inflacionaria. Esto es porque tal incertidumbre dificulta la toma de decisiones óptimas por parte de los agentes económicos.La evidencia empírica, para el periodo 1954:01-2002:08, apoya la hipótesis de que para el caso de Costa Rica mientras mayor es la inflación mayor es la incertidumbre respecto a esta variable. En los últimos siete años (1997-2002) la incertidumbre presenta la variación media más baja de todo el periodo. Además, se identifica un efecto asimétrico de la inflación sobre la incertidumbre inflacionaria, es decir, la incertidumbre inflacionaria tiende a incrementarse más para el siguiente periodo cuando la inflación pronosticada está por debajo de la inflación actual, que cuando la inflación pronosticada está por arriba de la tasa observada de inflación. Estos resultados tienen una clara implicación para la política monetaria. Para minimizar la dificultad que la inflación causa en la toma óptima de decisiones de los agentes económicos es necesario perseguir no solamente un nivel bajo de inflación sino que también sea estable.AbstractThis paper estimates a measure of inflationary uncertainty. An inflation model signals uncertainty when the forecast errors are heteroskedastic. By the specification of a GARCH (Generalized Autoregressive Conditional Heteroscedasticity) equation, for the variance of the error term of the inflation model, it is possible to estimate a proxy for inflationary uncertainty. By the simultaneous estimation of the inflation model and the GARCH equation, a new inflation model is obtained in which the forecast errors are homocedastic. Most economists agree that there is a positive correlation between inflationary uncertainty and the magnitude of the inflation rate, which, as was pointed out by Friedman (1977), represents one of costs associated with the persistence of inflation. This is because such uncertainty clouds the decision-making process of consumers and investors.The empirical evidence for the period 1954:01-2002:08 confirms that in the case of Costa Rica inflationary uncertainty increases as inflation rises. In the last seven years(1997-2002) the uncertainty present the mean variation most small of the period. In addition, inflation has an asymmetric effect on inflationary uncertainty. That is, when the inflation forecast is below the actual inflation, inflationary uncertainty increases for the next period. The opposite happens when the inflation forecast is above the observed rate of inflation. Besides, the absolute value of the change on uncertainty is greater in the first case than the second. These results have a clear implication for monetary policy. To minimize the disruptions that inflation causes to the economic decision-making process, it is necessary to pursue, not only a low level of inflation, but a stable one as well.

Volatility transmission between stock and exchange-rate markets: A connectedness analysis

This paper empirically investigates volatility transmission among stock and foreign exchange markets in seven major world economies during the period July 1988 to January 2015. To this end, we first perform a static and dynamic analysis to measure the total volatility connectedness in the entire period (the system-wide approach) using a framework recently proposed by Diebold and Yilmaz (2014). Second, we make use of a dynamic analysis to evaluate the net directional connectedness for each market. To gain further insights, we examine the time-varying behaviour of net pair-wise directional connectedness during the financial turmoil periods experienced in the sample period Our results suggest that slightly more than half of the total variance of the forecast errors is explained by shocks across markets rather than by idiosyncratic shocks. Furthermore, we find that volatility connectedness varies over time, with a surge during periods of increasing economic and financial instability.

Forecasting errors, directional accuracy and profitability of currency trading: The case of EUR/USD exchange rate

We provide a comprehensive study of out-of-sample forecasts for the EUR/USD exchange rate based on multivariate macroeconomic models and forecast combinations. We use profit maximization measures based on directional accuracy and trading strategies in addition to standard loss minimization measures. When comparing predictive accuracy and profit measures, data snooping bias free tests are used. The results indicate that forecast combinations, in particular those based on principal components of forecasts, help to improve over benchmark trading strategies, although the excess return per unit of deviation is limited.

Correcting the Errors : A Note on Volatility Forecast Evaluation Based on High-Frequency Data and Realized Volatilities

This note develops general model-free adjustment procedures for the calculation of unbiased volatility loss functions based on practically feasible realized volatility benchmarks. The procedures, which exploit the recent asymptotic distributional results in Barndorff-Nielsen and Shephard (2002a), are both easy to implement and highly accurate in empirically realistic situations. On properly accounting for the measurement errors in the volatility forecast evaluations reported in Andersen, Bollerslev, Diebold and Labys (2003), the adjustments result in markedly higher estimates for the true degree of return-volatility predictability.

Distributed energy resources scheduling considering real-time resources forecast

Energy resource scheduling is becoming increasingly important, such as the use of more distributed generators and electric vehicles connected to the distribution network. This paper proposes a methodology to be used by Virtual Power Players (VPPs), regarding the energy resource scheduling in smart grids and considering day-ahead, hour-ahead and realtime time horizons. This method considers that energy resources are managed by a VPP which establishes contracts with their owners. The full AC power flow calculation included in the model takes into account network constraints. In this paper, distribution function errors are used to simulate variations between time horizons, and to measure the performance of the proposed methodology. A 33-bus distribution network with large number of distributed resources is used.