Thesis (PhD)--Stellenbosch University, 1997. / One copy microfiche. / ENGLISH ABSTRACT: Discriminani analysis and logistic regression are techniques that can be used to classify entities of unknown origin into one of a number of groups. However, the underlying models and assumptions for application of the two techniques differ. In this study, the two techniques are compared with respect to classification of entities. Firstly, the two techniques were compared in situations where no data dependent variable selection took place. Several underlying distributions were studied: the normal distribution, the double exponential distribution and the lognormal distribution. The number of variables, sample sizes from the different groups and the correlation structure between the variables were varied to' obtain a large number of different configurations. .The cases of two and three groups were studied. The most important conclusions are: "for normal and double' exponential data linear discriminant analysis outperforms logistic regression, especially in cases where the ratio of the number of variables to the total sample size is large. For lognormal data, logistic regression should be preferred, except in cases where the ratio of the number of variables to the total sample size is large. " Variable selection is frequently the first step in statistical analyses. A large number of potenti8.Ily important variables are observed, and an optimal subset has to be selected for use in further analyses. Despite the fact that variable selection is often used, the influence of a selection step on further analyses of the same data, is often completely ignored. An important aim of this study was to develop new selection techniques for use in discriminant analysis and logistic regression. New estimators of the postselection error rate were also developed. A new selection technique, cross model validation (CMV) that can be applied both in discriminant analysis and logistic regression, was developed. ."This technique combines the selection of variables and the estimation of the post-selection error rate. It provides a method to determine the optimal model dimension, to select the variables for the final model and to estimate the post-selection error rate of the discriminant rule. An extensive Monte Carlo simulation study comparing the CMV technique to existing procedures in the literature, was undertaken. In general, this technique outperformed the other methods, especially with respect to the accuracy of estimating the post-selection error rate. Finally, pre-test type variable selection was considered. A pre-test estimation procedure was adapted for use as selection technique in linear discriminant analysis. In a simulation study, this technique was compared to CMV, and was found to perform well, especially with respect to correct selection. However, this technique is only valid for uncorrelated normal variables, and its applicability is therefore limited. A numerically intensive approach was used throughout the study, since the problems that were investigated are not amenable to an analytical approach. / AFRIKAANSE OPSOMMING: Lineere diskriminantanaliseen logistiese regressie is tegnieke wat gebruik kan word vir die Idassifikasie van items van onbekende oorsprong in een van 'n aantal groepe. Die agterliggende modelle en aannames vir die gebruik van die twee tegnieke is egter verskillend. In die studie is die twee tegnieke vergelyk ten opsigte van k1assifikasievan items. Eerstens is die twee tegnieke vergelyk in 'n apset waar daar geen data-afhanklike seleksie van veranderlikes plaasvind me. Verskeie onderliggende verdelings is bestudeer: die normaalverdeling, die dubbeleksponensiaal-verdeling,en die lognormaal verdeling. Die aantal veranderlikes, steekproefgroottes uit die onderskeie groepe en die korrelasiestruktuur tussen die veranderlikes is gevarieer om 'n groot aantal konfigurasies te verkry. Die geval van twee en drie groepe is bestudeer. Die belangrikste gevolgtrekkings wat op grond van die studie gemaak kan word is: vir normaal en dubbeleksponensiaal data vaar lineere diskriminantanalise beter as logistiese regressie, veral in gevalle waar die. verhouding van die aantal veranderlikes tot die totale steekproefgrootte groot is. In die geval van data uit 'n lognormaalverdeling, hehoort logistiese regressie die metode van keuse te wees, tensy die verhouding van die aantal veranderlikes tot die totale steekproefgrootte groot is. Veranderlike seleksie is dikwels die eerste stap in statistiese ontledings. 'n Groot aantal potensieel belangrike veranderlikes word waargeneem, en 'n subversamelingwat optimaal is, word gekies om in die verdere ontledings te gebruik. Ten spyte van die feit dat veranderlike seleksie dikwels gebruik word, word die invloed wat 'n seleksie-stap op verdere ontledings van dieselfde data. het, dikwels heeltemal geYgnoreer.'n Belangrike doelwit van die studie was om nuwe seleksietegniekete ontwikkel wat gebruik kan word in diskriminantanalise en logistiese regressie. Verder is ook aandag gegee aan ontwikkeling van beramers van die foutkoers van 'n diskriminantfunksie wat met geselekteerde veranderlikes gevorm word. 'n Nuwe seleksietegniek, kruis-model validasie (KMV) wat gebruik kan word vir die seleksie van veranderlikes in beide diskriminantanalise en logistiese regressie is ontwikkel. Hierdie tegniek hanteer die seleksie van veranderlikes en die beraming van die na-seleksie foutkoers in een stap, en verskaf 'n metode om die optimale modeldimensiete bepaal, die veranderlikes wat in die model bevat moet word te kies, en ook die na-seleksie foutkoers van die diskriminantfunksie te beraam. 'n Uitgebreide simulasiestudie waarin die voorgestelde KMV-tegniek met ander prosedures in die Iiteratuur. vergelyk is, is vir beide diskriminantanaliseen logistiese regressie ondemeem. In die algemeen het hierdie tegniek beter gevaar as die ander metodes wat beskou is, veral ten opsigte van die akkuraatheid waarmee die na-seleksie foutkoers beraam word. Ten slotte is daar ook aandag gegee aan voor-toets tipeseleksie. 'n Tegniek is ontwikkel wat gebruik maak van 'nvoor-toets berarningsmetode om veranderlikes vir insluiting in 'n lineere diskriminantfunksie te selekteer. Die tegniek ISin 'n simulasiestudie met die KMV-tegniek vergelyk, en vaar baie goed, veral t.o.v. korrekte seleksie. Hierdie tegniek is egter slegs geldig vir ongekorreleerde normaalveranderlikes, wat die gebruik darvan beperk. 'n Numeries intensiewe benadering is deurgaans in die studie gebruik. Dit is genoodsaak deur die feit dat die probleme wat ondersoek is, nie deur middel van 'n analitiese benadering hanteer kan word nie.

published_or_final_version / abstract / Economics and Finance / Master / Master of Philosophy

There are several methods of analysing time-to-event data. These include nonparametric approaches such as Kaplan-Meier estimation and parametric approaches such as regression modeling. Parametric regression modeling involves specifying the distribution of the survival time of the individuals, which are commonly chosen to be either exponential, Weibull, log- normal, log-logistic or gamma distributed. Another well known model that does not require assumptions about the hazard function to be made is the Cox proportional hazards model. However, there may be deviations from proportional hazards which may be explained by unaccounted random heterogeneity. In the early 1980s, a series of studies showed concern with the possible bias in the estimated treatment e®ect when important covariates are omitted. Other problems may be encountered with the traditional proportional hazards model when there is a possibility of correlated data, for instance when there is clustering. A method of handling these types of problems is by making use of frailty modeling. Frailty modeling is a method whereby a random e®ect is incorporated in the Cox pro- portional hazards model. While this concept is fairly simple to understand, the method of estimation of the ¯xed and random e®ects becomes complicated. Various methods have been explored by several authors, including the Expectation-Maximisation (EM) algorithm, pe- nalized partial likelihood approach, Markov Chain Monte Carlo (MCMC) methods, Monte Carlo EM approach and di®erent methods using Laplace approximation. The lack of available software is problematic for ¯tting frailty models. These models are usually computationally extensive and may have long processing times. However, frailty modeling is an important aspect to consider, particularly if the Cox proportional hazards model does not adequately describe the distribution of survival time. / Thesis (M.Sc.)-University of KwaZulu-Natal, Pietermaritzburg, 2006.

This study investigates the software automation and control framework for the MMT thermal system. Thermal-related effects on observing and telescope behavior have been considered during the entire software development process. Regression analysis of telescope and observatory subsystem data is used to characterize and model these thermal-related effects. The regression models help predict expected changes in focus and overall astronomical seeing that result from temperature variations within the telescope structure, within the primary mirror glass, and between the primary mirror glass and adjacent air (i.e., mirror seeing). This discussion is followed by a description of ongoing upgrades to the heating, ventilation and air conditioning (HVAC) system and the associated software controls. The improvements of the MMT thermal system have two objectives: 1) to provide air conditioning capabilities for the MMT facilities, and 2) to modernize and enhance the primary mirror (M1) ventilation system. The HVAC upgrade necessitates changes to the automation and control of the M1 ventilation system. The revised control system must factor in the additional requirements of the HVAC system, while still optimizing performance of the M1 ventilation system and the M1's optical behavior. An industry-standard HVAC communication and networking protocol, BACnet (Building Automation and Control network), has been adopted. Integration of the BACnet protocol into the existing software framework at the MMT is discussed. Performance of the existing automated system is evaluated and a preliminary upgraded automated control system is presented. Finally, user interfaces to the new HVAC system are discussed.

Military aircraft maintenance methods are moving from practices based on hard-time inspection and replacement intervals to one of Condition Based Maintenance (CBM). Benefits of CBM are the minimization of maintenance efforts and component replacement along with an increase in readiness and safety. Goodrich has developed the Integrated Mechanical Diagnostics Health and Usage Management System (IMD-HUMS) for the practices of CBM in helicopters. Great benefits have been realized with the IMD-HUMS system in regards to several maintenance practices, readiness, and safety. However, the total potential of the system in regards to these benefits for the multiple components observed by the IMDHUMS is not yet achieved. The IMD-HUMS gathers a great deal of pertinent, important data on the condition of multiple components and systems, but the meaning and full potential of all this data is not yet fully realized. The purpose of this research is to conduct and document a statistical analysis of IMD-HUMS produced data. Statistical applications of data mining, regression and classification trees are explored. The approaches used in the exploration of the IMD-HUMS acquired data sets are based on six electrical generators which displayed degradation or failure and hence required maintenance actions compared with sixty others which did not.

No description available.

Differential item functioning (DIF) detection rates were examined for the logistic regression and analysis of variance (ANOVA) DIF detection methods. The methods were applied to simulated data sets of varying test length (20, 40, and 60 items) and sample size (200, 400, and 600 examinees) for both equal and unequal underlying ability between groups as well as for both fixed and varying item discrimination parameters. Each test contained 5% uniform DIF items, 5% non-uniform DIF items, and 5% combination DIF (simultaneous uniform and non-uniform DIF) items. The factors were completely crossed, and each experiment was replicated 100 times. For both methods and all DIF types, a test length of 20 was sufficient for satisfactory DIF detection. The detection rate increased significantly with sample size for each method. With the ANOVA DIF method and uniform DIF, there was a difference in detection rates between discrimination parameter types, which favored varying discrimination and decreased with increased sample size. The detection rate of non-uniform DIF using the ANOVA DIF method was higher with fixed discrimination parameters than with varying discrimination parameters when relative underlying ability was unequal. In the combination DIF case, there was a three-way interaction among the experimental factors discrimination type, relative ability, and sample size for both detection methods. The error rate for the ANOVA DIF detection method decreased as test length increased and increased as sample size increased. For both methods, the error rate was slightly higher with varying discrimination parameters than with fixed. For logistic regression, the error rate increased with sample size when relative underlying ability was unequal between groups. The logistic regression method detected uniform and non-uniform DIF at a higher rate than the ANOVA DIF method. Because the type of DIF present in real data is rarely known, the logistic regression method is recommended for most cases.

This study aims to investigate the determinants of capital structure in small and medium-sized enterprises (SMEs) in Malaysia and their effect on firms’ performance. The study addresses the following primary question: What are the factors that influence the capital structure of SMEs in Malaysia? The sample of this research is SMEs in the east coast region of Malaysia. Adopting a positivist paradigm, the research design includes a preliminary study comprising 25 interviews with the owner-managers of SMEs, which is analysed using thematic analysis. The results are used to finalise the conceptual framework for the main study, which takes the form of a self-completion questionnaire survey. Usable responses were received from 384 firms, giving a response rate of 75.3%. The survey data is analysed using a series of binomial logistic regression models. Results reveal that there was no indication for the impact of owner’s education and experience on capital structure decisions. Other owner-related factors, firm characteristics, management performance and environment were found to relate to all types of capital structure. Both complete and partial mediating effects are also discovered in this study. The results provide evidence to support the pecking order hypothesis (Myers, 1984; Myers and Majluf, 1984), agency theory (Jensen and Meckling, 1976) and culture model of Schwartz (1994). It appeared that owner-managers in Malaysia do not strive to adjust their capital structure towards some optimal debt ratio, which is contrary to the static trade-off theory (DeAngelo and Masulis, 1980) of capital structure. This study makes several important contributions to the existing studies of capital structure. This research led to the development of a model of capital structure determinants by integrating factors related to owner-managers, firms, culture, and environment. This study incorporates methodological triangulation that may mitigate the problem of the difficulties in accessing financial data of SMEs in Malaysia. This study also provides meaningful insight into the financing preferences of the owner-managers with relevant implementations to academics, business practitioners, financial providers and policymakers. The research findings should assist owner-managers in making optimal capital structure decisions as well as help the policymaker in making an appropriate policy on the financing.

Artificial Neural Networks (ANNs) are powerful predictors, however, they essentially function like 'black boxes' because they lack explanatory power. Various algorithms have been developed to examine input influences and interactions thus enhancing understanding of the function being modeled. The study of facial attractiveness is one domain that could potentially benefit from ANN models. The literature shows that the relationship between attractiveness and facial attributes is complex and not yet fully understood. In this project, a feed-forward ANN was trained with backpropagation to 0.86 classification using 8-fold cross validation. The dataset consisted of 88 female facial images, each containing 17 geofacial measurements, a random noise variable, and a rating. Input 'clamping' and the Connection Weight Approach (Olden & Jackson, 2002), were implemented and the results were examined in terms of the facial attractiveness domain. In general, the results suggest that more feminized and asymmetrical features enhance facial attractiveness.

This study examined the influences of attachment style and level of optimism on career self-efficacy. It was hypothesized that level of optimism moderates the relationship between attachment style and career self-efficacy. Participants were 173 college students who completed the Experiences in Close Relationships (ECR; K.A. Brennan, C.L. Clark, & P.R. Shaver, 1998), the Life Orientation Test - Revised (LOT-R; M. Sheier, C. Carver, & M. Bridges, 1994), and the Career Decision-Making Self-Efficacy - Short Form (CDMSE-SF; N. Betz, K. Klein, & K. Taylor, 1996). The moderator model was not supported, but hierarchical regression revealed a positive relationship between optimism and career self-efficacy. Analyses also revealed a negative correlation between avoidant attachment style and career self-efficacy.