Wong Hing-Kwok. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2003. / Includes bibliographical references (leaves 34-35). / Abstracts in English and Chinese. / Abstract --- p.i / Acknowledgement --- p.iii / Chapter 1 --- Introduction --- p.1 / Chapter 2 --- Radial Basis Function (RBF) of Neural Network --- p.5 / Chapter 2.1 --- Neural Network --- p.6 / Chapter 2.2 --- Radial Basis Function (RBF) Network --- p.8 / Chapter 2.3 --- Model Specification --- p.10 / Chapter 2.4 --- Estimation --- p.12 / Chapter 3 --- RBF in Performance Attribution --- p.17 / Chapter 3.1 --- Background of Data Set --- p.18 / Chapter 3.2 --- Portfolio Construction --- p.20 / Chapter 3.3 --- Portfolio Rebalance --- p.22 / Chapter 3.4 --- Result --- p.23 / Chapter 4 --- Comparison --- p.26 / Chapter 4.1 --- Standard Linear Model --- p.27 / Chapter 4.2 --- Fixed Additive Model --- p.28 / Chapter 4.3 --- Refined Additive Model --- p.29 / Chapter 4.4 --- Result --- p.30 / Chapter 5 --- Conclusion --- p.32 / Bibliography --- p.34

Shum Chun-Yin. / Thesis (M.Phil.)--Chinese University of Hong Kong, 2003. / Includes bibliographical references (leaves 63-64). / Abstracts in English and Chinese. / Abstract --- p.i / Acknowledgement --- p.iv / Chapter 1 --- Introduction --- p.1 / Chapter 2 --- Ordinal Contingency Table --- p.5 / Chapter 2.1 --- Model --- p.5 / Chapter 2.2 --- The Maximum Likelihood Method --- p.7 / Chapter 2.3 --- Limitation of the Maximum Likelihood Estimation in Large Sample --- p.8 / Chapter 2.4 --- The Partition Maximum Likelihood Approach --- p.9 / Chapter 3 --- Modification of the Partition Maximum Likelihood Approach --- p.12 / Chapter 3.1 --- The Modified Partition Maximum Likelihood Approach --- p.12 / Chapter 3.2 --- Mx Implementation --- p.14 / Chapter 3.2.1 --- Maximum Likelihood Procedure --- p.14 / Chapter 3.2.2 --- Modified PML Procedure --- p.15 / Chapter 3.3 --- Examples --- p.16 / Chapter 3.3.1 --- Example 1 : Attitudes of Morality and Equality --- p.16 / Chapter 3.3.2 --- Example 2 : A Panel Model for Political Efficacy --- p.17 / Chapter 3.4 --- Limitation of the Modified PML Approach --- p.19 / Chapter 3.5 --- Simulation Study for the Modified PML Approach --- p.20 / Chapter 4 --- Generalization to Structural Equation Model --- p.22 / Chapter 4.1 --- Model --- p.23 / Chapter 4.2 --- Procedure --- p.24 / Chapter 4.3 --- Examples --- p.26 / Chapter 4.3.1 --- Example 1 : Attitudes of Morality and Equality --- p.26 / Chapter 4.3.2 --- Example 2 : A Panel Model for Political Efficacy --- p.28 / Chapter 5 --- Generalization to Stochastic Constraints on Thresholds --- p.31 / Chapter 5.1 --- Model --- p.32 / Chapter 5.2 --- Bayesian Analysis of the Model --- p.33 / Chapter 5.3 --- Examples --- p.35 / Chapter 5.3.1 --- Example 1 : Attitudes of Morality and Equality --- p.35 / Chapter 5.3.2 --- Example 2 : A Panel Model for Political Efficacy --- p.36 / Chapter 6 --- Conclusion and Discussion --- p.38 / Chapter A --- Mx Script of the ML Estimation - for Example 1 --- p.40 / Chapter B --- Mx Script of the Modified PML Estimation - for Example 1 --- p.42 / Chapter C --- Mx Script of the Modified PML Estimation - for Example 2 --- p.45 / Bibliography --- p.63

No description available.

No description available.

Diversity indices confound two components: species richness and evenness. Community structure should therefore be evaluated by employing separate measures of the number of species and their relative abundances. However, the relative abundances of species are dependent on the abundance measure used. Although the use of biomass or productivity is recommended by theory, in practice a surrogate measure is more often used. Frequency (local or relative) and point-quadrat cover provide two objective measures of abundance which are fast, less destructive and avoid problems associated with distinguishing individuals. However, both give discrete bounded data which may further alter the relative abundances of species. These measures have a long history of use and, as the need for objective information on biodiversity becomes more pressing, their use is likely to become more widespread. Consequently, it seems appropriate to investigate the effect of these abundance measures, and the resolution at which they are used, on calculated evenness. Field, artificial and simulated data were used to investigate the effect of abundance measure and resolution on evidence for community structure. The field data consisted of seventeen sites. Sites from four vegetation types (saltmeadow, geothermal, ultramafic and high-altitude meadow) were sampled in three biogeographical regions. Most of the indices of community structure (species richness, diversity and evenness) detected differences between the different vegetation types, and different niche-apportionment models were fitted to the field data from saltmeadow and geothermal vegetation. Estimates of community structure based on local frequency and point-quadrat data differed. Local frequency tended to give higher calculated evenness; whereas point-quadrat data tended to fit to niche apportionment models where local frequency data failed. The effect of resolution on the eighteen evenness indices investigated depended on community species richness and the particular index used. The investigated evenness indices were divided into three groups (symmetric, continuous and traditional indices) based on how they ranked real and artificially constructed communities. Contrary to Smith and Wilson�s recommendation the symmetric indices E[VAR] and E[Q] proved unsuitable for use with most types of plant data. In particular, E[Q] tends to assign most communities low values and has a dubious relationship with intrinsic evenness. The continuous indices, E[MS] and E[2,1], were the indices best able to discriminate between field, artificial and simulated communities, and their use should be re-evaluated. Traditional indices used with low resolution tended to elevate the calculated evenness, especially in species-rich communities. The relativized indices, E[Hurlbert] and EO[dis], were an exception, as they were always able to attain the minimum of zero; however, they were more sensitive to changes in resolution, particularly when resolution was low. Overall, traditional indices based on Hill�s ratios, including E[1/D] (=E[2,0]), and G[2,1] gave the best performance, while the general criticism of the use of Pielou�s J� as an index of evenness was further substantiated by this study. As a final recommendation, ecologists are implored to investigate their data and the likely effects that sampling and analysis have had on the calculated values of their indices.

Multiple linear regression was used to develop equations for 12-, 24-, and 36-hour surface wind forecasts for the wind energy site at Goodnoe Hills. Equations were derived separately for warm and cool seasons. The potential predictors included LFM II model output, MOS surface wind forecasts extrapolated from surrounding stations, pressure observations corrected to mean sea level, and two types of climatological variables. Forecasts of wind speed and direction were formulated for an independent sample of predictands and predictors. The forecasts were evaluated using standard methods of forecast verification and the results are summarized in terms of several verification scores. Comparisons of scores were made by season, projection time, and cycle (or preparation) time, and some patterns were evident in the scores with respect to these stratifications. The minimum value of the mean absolute error attained by the forecast system presented here was 5.64 mph for a 12-hour, cool season forecast equation. The minimum value of the root mean square error was 7.57 mph for a 12-hour, warm season forecast equation. Comparison of these results with the results of other statistical wind forecasting studies indicates that the forecast equations for Goodnoe Hills are of comparable accuracy to the equations developed for other wind energy sites. Suggestions for future investigations of statistical wind forecasting are offered as well as recommendations concerning ways of improving the forecasting system described in this study. / Graduation date: 1983

Statisticians often focus on sampling or experimental design and data analysis while paying less attention to how the response is measured. However, the ideas of statistics may be applied to measurement problems with fruitful results. By examining the errors of measured responses, we may gain insight into the limitations of current measures and develop a better understanding of how to interpret and qualify the results. The first chapter considers the problem of measuring taxonomic richness as an index of habitat quality and stream health. In particular, we investigate numerical taxa richness (NTR), or the number of observed taxa in a fixed-count, as a means to assess differences in taxonomic composition and reduce cost. Because the number of observed taxa increases with the number of individuals counted, rare taxa are often excluded from NTR with smaller counts. NTR measures based on different counts effectively assess different levels of rarity, and hence target different parameters. Determining the target parameter that NTR is "really" estimating is an important step toward facilitating fair comparisons based on different sized samples. Our first study approximates the parameter unbiasedly estimated by NTR and explores alternatives for estimation based on smaller and larger counts. The second investigation considers response error resulting from panel evaluations. Because people function as the measurement instrument, responses are particularly susceptible to variation not directly related to the experimental unit. As a result, observed differences may not accurately reflect real differences in the products being measured. Chapter Two offers several linear models to describe measurement error resulting from unanchored responses across successive evaluations over time, which we call u-errors. We examine changes to Type I and Type II error probabilities for standard F-tests in balanced factorial models where u-errors are confounded with an effect under investigation. We offer a relatively simple method for determining whether or not distributions of mean square ratios for testing fixed effects change in the presence of u-error. In addition, the validity of the test is shown to depend both on the level of confounding and whether not u-errors vary about a nonzero mean. / Graduation date: 2002

Many ecological populations can be interpreted as response surfaces; the spatial patterns of the population vary in response to changes in the spatial patterns of environmental explanatory variables. Collection of a probability sample from the population provides unbiased estimates of the population parameters using design based estimation. When information is available for the environmental explanatory variables, model based procedures are available that provide more precise estimates of population parameters in some cases. In practice, not all of these environmental explanatory variables will be known. When the spatial coordinates of the population units are available, a spatial model can be used as a surrogate for the unknown, spatially patterned explanatory variables. Design based and model based procedures will be compared for estimating parameters of the population of Acid Neutralizing Capacity (ANC) of lakes in the Adirondack Mountains in New York. Results from the analysis of this population will be used to elucidate some general principles for model based estimation of parameters of spatial populations. Results indicate that using model based estimates of population parameters provide more precise estimates than design based estimates in some cases. In addition, including spatial information as a surrogate for spatially patterned missing covariates improves the precision of the estimates in some cases, the degree to which depends upon the model chosen to represent the spatial pattern. When the probability sample is selected from the spatial population is a stratified sample, differences in stratum variances need to be accounted for when residual spatial covariance estimation is desired for the entire population. This can be accomplished by scaling residuals by their estimated stratum standard deviation functions, and calculating the residual covariance using these scaled residuals. Results here demonstrate that the form of scaling influences the estimated strength of the residual correlation and the estimated correlation range. / Graduation date: 1997

Product development efforts are extremely important to a company's success in today's global competitive business environment. Yet, these highly consequential efforts are terribly nebulous to a point that past experiences are inherently underutilized. This thesis demonstrates a methodology to quantify past product development efforts in an attempt to better utilize past experiences. The methodology is centered around conducting an observational study, using regression analysis to expose relationships between various aspects of past product development efforts. In the study, products developed in the past serve as observational units, various cost and time variables serve as dependent variables, and a variety of variables characterizing product development efforts serve as independent variables. The nominal group technique is employed, along with formal personal interviewing, to identify the many different variables targeted for data collection. Regression analysis is used to test and identify relationships between the multitude of combinations of dependent and independent variables. Three simple model forms are used to 'capture' any potential relationship: a straight line model, an exponential model, and a natural-logarithmic model. Dependent-independent variable combinations that have met a given statistical criteria, in one of these three model forms, are labeled statistically notable, and later classified as practically relevant. The applicability of the methodology is demonstrated by presenting 'generic' results obtained by making use of information and historical data from a well established electronics company who wishes to be referred to as Company X. These results show that cost and/or time increase with the increase in: the number of parts in a product, the number of assembly processes, quality/utility of the product, or a product performance enabling specification. Furthermore, product shape is found to be associated with cost and time. Interestingly though, only a relatively few variables were found to be associated with time as compared to cost. The statistical models that were identified can serve as a quantitative historical record and perhaps a prediction tool for Company X, giving them a competitive advantage in their future product development efforts. / Graduation date: 1997

Graduation date: 1994