Please read the abstract in the section 00front of this document / Dissertation (MSc (Mathematics of Finance))--University of Pretoria, 2006. / Mathematics and Applied Mathematics / unrestricted

Report to Hydrology Laboratory, Office of Hydrology, National Weather Service, NOAH, Dept. of Commerce, Contract 6 -35229 / The value of a forecast system in preventing urban property damage depends on the accuracy of the forecasts, the time at which they are received, the response by the floodplain dweller and the êfficacy of that response. A systems model of the overall flood forecast -response system is developed. Evaluation of the system is accomplished by a decision theoretic methodology. A case study is done for Milton, Pennsylvania, which evaluates the present system and potential changes to it. It is concluded that the sequential nature of the forecast sequence must be considered in modeling the flood forecast -response system if a meaningful evaluation of the economic value of the system is to be obtained. Methodology for obtaining the parameterization of the model from the available data is given. Computer programs have been written to handle a good portion of the calculations. While more work is needed on obtaining accurate parameterization of certain parts of the model, such as the actual response to forecasts; use of the procedures and programs as they now stand produces reasonable evaluations.

This study investigated forecasting accuracy over time. Several quantitative and qualitative forecasting models were tested and a number of combinational methods was investigated. Six time series methods, one causal model, and one subjective technique were compared in this study. Six combinational forecasts were generated and compared to individual forecasts. A combining technique was developed. Thirty data sets, obtained from a market leader in the cosmetics industry, were used to forecast sales. All series represent monthly sales from January 1985 to December 1989. Gross sales forecasts from January 1988 to June 1989 were generated by the company using econometric models. All data sets exhibited seasonality and trend.

The Madden-Julian oscillation (MJO) and the quasi-biennial oscillation (QBO) are two key modes of variability in the tropical atmosphere. The MJO, characterized by propagating, planetary-scale signals in convection and winds, is the main source of subseasonal variability and predictability in the tropics. The QBO is a ~28-month cycle in which the tropical stratospheric zonal winds alternate between easterly and westerly regimes. Via thermal wind balance these winds induce temperature anomalies, and both wind and temperature signals reach the tropopause. Recent observational results show a remarkably strong link between the MJO and the QBO during boreal winter: the MJO is stronger and more predictable when QBO winds in the lower stratosphere are easterly than when winds are westerly. Despite its important implications for MJO theory and prediction, the physical processes driving the MJO-QBO interaction are not well-understood. In this thesis, we use a hierarchy of models – including a cloud-resolving model, a forecast model, and a global climate model – to examine whether models can reproduce the MJO-QBO link, and better understand the possible mechanisms driving the connection. Based in part on our modeling findings, we further explore observed QBO temperature signals thought to be important for the MJO-QBO link. After providing necessary background and context in the first two chapters, the third chapter looks at the MJO-QBO link in a small-domain, cloud-resolving model. The model successfully simulates convection associated with two MJO events that occurred during the DYNAMO field campaign. To examine the effect of QBO, we add various QBO temperature and wind anomalies into the model. We find that QBO temperature anomalies alone, without wind anomalies, qualitatively affect the model MJO similarly to the observed MJO-QBO connection. QBO wind anomalies have no clear effect on the modeled MJO. We note however that the MJO response is quite sensitive to the vertical structure of the QBO temperature anomalies, and for realistic temperature signals the model response is very small. In the fourth chapter, we look at the MJO-QBO link in a state-of-the-art global forecast model with a good representation of the MJO. We conduct 84 hind-cast experiments initialized on dates across winters from 1989-2017. For each of these dates, we artificially impose an easterly and a westerly QBO in the stratospheric initial conditions, and examine the resulting changes to the simulated MJO under different stratospheric states. We find that the effect of the QBO on the model MJO is of the same sign as observations, but is much smaller. A large sample size is required to capture any QBO signal, and tropospheric initial conditions seem more important than the stratosphere in determining the behavior of the simulated MJO. Despite the weak signal, we find that simulations with stronger QBO temperature anomalies have a stronger MJO response. In the fifth chapter, we conduct experiments in recent versions of a NASA general circulation model. We find that a version with a high vertical resolution generates a reasonable QBO and MJO, but has no MJO-QBO link. However, this model has weaker-than-observed QBO temperature anomalies, which may explain the lack of an MJO impact. To explore this potential bias, we impose the QBO by nudging the model stratospheric winds towards reanalysis, leading to more realistic simulation of QBO temperature anomalies. Despite this, the model still fails to show a strong MJO-QBO link across several ensemble experiments and sensitivity tests. We conclude with discussion of possible reasons why the model fails to capture the MJO-QBO connection. The sixth chapter examines QBO temperature signals in a range of observational and reanalysis datasets. In particular, we are motivated by two elements of the MJO-QBO relationship which are especially puzzling: the seasonality (i.e. that the MJO-QBO link is only significant in boreal winter) and long-term trend (i.e. that the MJO-QBO link seems to have only emerged since the 1980s). By examining QBO temperature signals around the tropopause, we highlight changes to the strength and structure of QBO temperature anomalies both in boreal winter and in recent decades. Whether these changes are linked to the MJO-QBO relationship, and what more generally might explain them, is not presently clear. Overall, we demonstrate that capturing the MJO-QBO relationship in a variety of models is a difficult task. The majority of evidence indicates that QBO-induced temperature anomalies are a plausible pathway through which the QBO might modulate the MJO, but the theoretical description of precisely how these temperature anomalies may impact convection is lacking and likely more nuanced than the literature to date suggests. Most models show only a weak modulation of the MJO associated with changes in upper-tropospheric temperatures, and even when those temperature signals are artificially enhanced, comprehensive GCMs still fail to show a significant MJO-QBO connection. Our observational study indicates that temperature anomalies associated with the QBO show striking modulations on various timescales of relevance to the MJO-QBO link, but do not conclusively demonstrate a clear connection to the MJO. This difficulty simulating a strong MJO-QBO connection suggests that models may lack a key process in driving the MJO and coupling the tropical stratosphere and troposphere. It is further possible that the observed link may be in some regards different than is currently theorized -- for example statistically not robust, due to non-stratospheric processes, or driven by some mechanism that has not been suitably explored.

published_or_final_version / Business Administration / Master / Master of Business Administration

published_or_final_version / Statistics and Actuarial Science / Master / Master of Philosophy

This thesis examined the relationship between basic employment and derived employment for the Muncie Standard Metropolitan Statistical Area. The study also presented graphically, by use of monthly time series data from 1965 to 1972, employment growth patterns by industrial group. A conceptual model was developed relating the contributions of, basic employment to the growth of derived employment. The model also related the lagged, primary -secondary, and wage scale effects of basic employment upon derived employment. The model assumed that the earnings off workers in basic employment is a major determinant of derived employment growth. Total monthly earnings for each industrial group were specified as explanatory variables in a series of multiple regression equations to determine the various basic industries' contributions to the growth of derived employment. Five-year derived employment projections were computed assuming various growth rates in earnings for each of the significant basic industries. The study concluded that the growth of Ball State University employment (basic government) was the most significant factor effecting the growth of derived employment in the Muncie Standard Metropolitan Statistical Area.

A research report submitted to the Faculty of Commerce, Law and Management, University of the Witwatersrand, Johannesburg, in partial fulfillment of the requirements for the degree of Master of Management in Finance and Investment. Johannesburg, 2016 / The prediction of stock price and index level in a financial market is an interesting but highly complex and intricate topic. Advancements in prediction models leading to even a slight increase in performance can be very profitable. The number of studies investigating models in predicting actual levels of stocks and indices however, far exceed those predicting the direction of stocks and indices. This study evaluates the performance of ensemble prediction models in predicting the daily direction of the JSE All-Share index. The ensemble prediction models are benchmarked against three common prediction models in the domain of financial data prediction namely, support vector machines, logistic regression and k-nearest neighbour. The results indicate that the Boosted algorithm of the ensemble prediction model is able to predict the index direction the best, followed by k-nearest neighbour, logistic regression and support vector machines respectively. The study suggests that ensemble models be considered in all stock price and index prediction applications. / MT2017

RESEARCH THESIS SUBMITTED TO THE FACULTY OF COMMERCE, LAW & MANAGEMENT IN PARTIAL FULLFILMENT OF THE REQUIREMENTS OF THE MASTER OF MANAGEMENT IN FINANCE & INVESTMENTS DEGREE UNIVERSITY OF THE WITWATERSRAND JOHANNESBURG February, 2016 / In this study we examine the relationship between international oil prices and the South African Rand/US Dollar exchange rate. We also determine the direction of causality between these two variables. We further ascertain the magnitude of the influence of oil prices to the exchange rate compared to other theoretically driven macroeconomic variables. A forecasting exercise is also undertaken to determine whether oil prices contain information about future Rand/Dollar exchange rate. Drawing from the works of Aliyu (2009) and Jin (2008) we use VAR based cointegration technique and vector error correction model (VECM) for the long run and short run analysis respectively. The results show that there is a unidirectional causality running from oil prices to exchange rate and not the other way round. We also find that a 1% permanent increase in oil prices results in 0.17% appreciation of the Rand against the US Dollar; a 1% permanent increase in money aggregates results in 21.3% depreciation of the Rand and a 1% increase in business cycles results in 0.29% depreciation of the Rand in the long run. A 1% increase in inflation and interest rates is found to result in a 0.09% and 0.005% depreciation on the Rand respectively. Our short run analysis indicates that 4.4% of the Rand/Dollar exchange rate disequilibrium can be corrected within a month. Oil prices are found to contain some information about the future Rand/US Dollar exchange rate when the VAR model is used for forecasting. This study has shown there is a causal relationship between oil prices and the strength of the Rand against the Dollar and, therefore, recommends diversification of the economy and more use of green energy. Strategies to reduce capital flight and trade-related capital is also recommended by this study. Key Words: Exchange rate, Oil price, forecasting, vector autoregressive (VAR) model, cointegration, vector error correction model (VECM), causality / MT2016

Classification algorithms represent a rich set of tools, which train a classification model from a given training and test set, to classify previously unseen test instances. Although existing methods have studied classification algorithm performance with respect to feature selection, noise condition, and sample distributions, our existing studies have not addressed an important issue on the classification algorithm performance relating to feature deletion and addition. In this thesis, we carry out sensitive study of classification algorithms by using feature deletion and addition. Three types of classifiers: (1) weak classifiers; (2) generic and strong classifiers; and (3) ensemble classifiers are validated on three types of data (1) feature dimension data, (2) gene expression data and (3) biomedical document data. In the experiments, we continuously add redundant features to the training and test set in order to observe the classification algorithm performance, and also continuously remove features to find the performance of the underlying classifiers. Our studies draw a number of important findings, which will help data mining and machine learning community under the genuine performance of common classification algorithms on real-world data. / Includes bibliography. / Thesis (M.S.)--Florida Atlantic University, 2014. / FAU Electronic Theses and Dissertations Collection