None

In this thesis, we are interested in empirical likelihood (EL) methods for two-sample problems, with focus on the difference of the two population means. A weighted empirical likelihood method (WEL) for two-sample problems is developed. We also consider a scenario where sample data on auxiliary variables are fully observed for both samples but values of the response variable are subject to missingness. We develop an adjusted empirical likelihood method for inference of the difference of the two population means for this scenario where missing values are handled by a regression imputation method. Bootstrap calibration for WEL is also developed. Simulation studies are conducted to evaluate the performance of naive EL, WEL and WEL with bootstrap calibration (BWEL) with comparison to the usual two-sample t-test in terms of power of the tests and coverage accuracies. Simulation for the adjusted EL for the linear regression model with missing data is also conducted.

Cluster analysis has been identified as a core task in data mining for which many different algorithms have been proposed. The diversity, on one hand, provides us a wide collection of tools. On the other hand, the profusion of options easily causes confusion. Given a particular task, users do not know which algorithm is good since it is not clear how clustering algorithms should be evaluated. As a consequence, users often select clustering algorithm in a very adhoc manner. A major challenge in evaluating clustering algorithms is the scarcity of real data with a "correct" ground truth clustering. This is in stark contrast to the situation for classification tasks, where there are abundantly many data sets labeled with their correct classifications. As a result, clustering research often relies on labeled data to evaluate and compare the results of clustering algorithms. We present a new perspective on how to use labeled data for evaluating clustering algorithms, and develop an approach for comparing clustering algorithms on the basis of classification labeled data. We then use this approach to support a novel technique for choosing among clustering algorithms when no labels are available. We use these tools to demonstrate that the utility of an algorithm depends on the specific clustering task. Investigating a set of common clustering algorithms, we demonstrate that there are cases where each one of them outputs better clusterings. In contrast to the current trend of looking for a superior clustering algorithm, our findings demonstrate the need for a variety of different clustering algorithms.

No description available.

Mechanistic-Empirical (M-E) methods are the cornerstone of current pavement engineering practice because of their enhanced predicting capabilities. Such predicting power demands richer input data, computational power, and calibration of the empirical components against distress measurements in the field. In an effort to spearhead the transition to M-E design in Uruguay, the aim of this Project is twofold: (1) develop an open-source, MEPDG-based, simplified M-E tool for Uruguayan flexible pavements [Product-One], and (2) compile a library of Uruguayan input data for design [Product-Two]. A functional, Matlab-based beta version of Product-One with default calibration parameters and a first collection of Uruguayan input data are presented herein. The Product-One beta is capable of designing hot-mix asphalt (HMA) structures over granular bases on top of the subgrade. Product-Two features climate information from the INIA weather station network, traffic distribution patterns for select Uruguayan highways, standard-based (Level-3) HMA properties, and Level-3 and Level-2 unbound materials' parameters. Product-One's outcomes were against other available M-E software, as a means to test the code's performance: Product-One reported a distress growth similar to CR-ME (MEPDG-based) on default calibration parameters but different to MeDiNa (calibrated core). In conclusion, Product-One managed to perform like another MEPDG-based software under the same design inputs and constraints, accomplishing one of this Thesis' objectives. However, Product-Two could not be created to the initially-desired extent. Nevertheless, the author remains confident that significant leaps forward can be made with little extra effort and further research on M-E design can be encouraged from this project. / Master of Science / The design of pavement structures historically relied on methodologies developed after experimental results, the so-called “empirical methods”. Advances in technology over the recent years allowed for more complex but more reliable methods – the mechanistic-empirical (M-E) methods – to finally be adopted by practitioners both in the US and abroad. In an effort to encourage the transition to M-E design in Uruguay, this project aimed at developing an open-source M-E design tool for Uruguayan flexible pavements based on the American MEPDG design methodology [Product-One], and assembling a library of Uruguayan data necessary for design with such an M-E method [Product-Two]. In this project, a beta version of the Product-One design tool for the design of asphalt-surfaced pavements and a collection of climate, traffic distribution, and materials’ properties data from Uruguayan sources for design is presented (load information was not available for this project); this Thesis is the log of the data collection effort as well as the guide to using and understanding all the components of Product-One. In addition, Product-One has been tested by comparing its pavement design results for a given Uruguayan highway against other M-E design software tools: MeDiNa and CR-ME. Product-One’s outcomes resembled the results given by CR-ME (also MEPDG-based) but differed with those from MeDiNa (crafted specifically for design of Brazilian roads). In conclusion, Product-One managed to perform like another MEPDG-based design tool under the same inputs and constraints, accomplishing one of this Thesis’ objectives. However, some of the Uruguayan information sought for could not be retrieved and so added to Product-Two. Anyway, the author remains confident that both Products can be significantly improved with little extra effort and that this project may encourage further research on M-E design.

The use of doubly robust estimators is a key for estimating the population mean response in the presence of incomplete data. Cao et al. (2009) proposed an alternative doubly robust estimator which exhibits strong performance compared to existing estimation methods. In this thesis, we apply the jackknife empirical likelihood, the jackknife empirical likelihood with nuisance parameters, the profile empirical likelihood, and an empirical likelihood method based on the influence function to make an inference for the population mean. We use these methods to construct confidence intervals for the population mean, and compare the coverage probabilities and interval lengths using both the ``usual'' doubly robust estimator and the alternative estimator proposed by Cao et al. (2009). An extensive simulation study is carried out to compare the different methods. Finally, the proposed methods are applied to two real data sets.

Pietra ratio (Pietra index), also known as Robin Hood index, Schutz coefficient (Ricci-Schutz index) or half the relative mean deviation, is a good measure of statistical heterogeneity in the context of positive-valued data sets. In this thesis, two novel methods namely "adjusted jackknife empirical likelihood" and "extended jackknife empirical likelihood" are developed from the jackknife empirical likelihood method to obtain interval estimation of the Pietra ratio of a population. The performance of the two novel methods are compared with the jackknife empirical likelihood method, the normal approximation method and two bootstrap methods (the percentile bootstrap method and the bias corrected and accelerated bootstrap method). Simulation results indicate that under both symmetric and skewed distributions, especially when the sample is small, the extended jackknife empirical likelihood method gives the best performance among the six methods in terms of the coverage probabilities and interval lengths of the confidence interval of Pietra ratio; when the sample size is over 20, the adjusted jackknife empirical likelihood method performs better than the other methods, except the extended jackknife empirical likelihood method. Furthermore, several real data sets are used to illustrate the proposed methods.

Neste trabalho aplica-se um método semi-empírico que utiliza uma técnica de otimização não linear para determinação das eficiências volumétrica e combinada do compressor hermético alternativo. Relações para estimar aproximadamente o fluxo de massa e a potência elétrica do compressor também são propostas. Todas estas características do compressor são calculadas através das relações físicas do modelo, empregadas nos cálculos de otimização. O método é implementado no software EES (Engineering Equation Solver) e baseia-se nos trabalhos de Jahing (1999) e Jahing et al. (2000). No presente método, cinco medições experimentais do fluxo de massa e potência elétrica são suficientes para determinar os parâmetros de ajuste do modelo. Este procedimento permite a geração de mapas de compressores satisfatórios sem a necessidade de um maior número de dados experimentais como no caso da norma ARI 540. Estes resultados obtidos com o modelo podem ser usados para o projeto de novos compressores. / In the present work is applied a semi-empirical method that uses a non-linear optimization technique for determination of the volumetric and combined efficiencies of hermetic reciprocating compressor. Relations to approximately estimate the mass flow and the electric power of the compressor are also proposed. All these compressor characteristics are calculated through physical model relations, used in the optimization calculations. The method is implemented in the EES (Engineering Equation Solver) software and is based on the works of Jahing (1999) and Jahing et al. (2000). In the method, four experimental measurements of the mass flow and electric power are enough to determine the fitting parameters of the model. This procedure allows the generation of satisfactory compressor maps without the necessity of a higher number of experimental data, as in the case of norm ARI 540 application. The results obtained with the model can be used for the design of new compressors.

This research examines the different social costs that have a bearing on the form and content of regulatory intervention underpinning the Auto-Enrolment (AE) pensions’ regime and whether these are recognised within current UK legislation. AE requires employers to assess the workforce and, subject to qualifying criteria, enrol them into a pension and make contributions to the pension. Workers are also required to contribute to the pension although they can choose to leave the scheme, Since its introduction in 2012, more than 4.7 million workers in the UK have now been enrolled into a pension which they did not choose highlighting the impact of the regulatory intervention. The starting point for my research is to consider the history of pension legislation in the UK and the circumstances leading up to the introduction of AE. The next step is to consider whether this legislative intervention is paternalistic and, if so, whether justified. I then consider whether the implementing legislation functions as it was intended or whether there are any unintended consequences. Parliamentary debate at the inception of AE illustrates an intention to ensure that individuals save more for themselves for their own good. Behavioural economics was widely referred to as authority on the hypothesis that individuals do not make rational financial decisions for themselves, justifying legislative intervention to ensure that private pension saving is the default position for all workers in the UK. Paternalistic legislation can be difficult to justify where individual choice is overridden entirely and so the legislation allows for individuals to opt out – soft paternalism or nudge. There is an assumption that low opt-out rates indicate that individuals now agree that the pension saving is in their best interests, justifying the intervention. However, there has been a marked dearth of research into AE savers' actual perceptions of how the schemes work for them. My research has come to close this gap and also allows drawing up some normative conclusions regarding the proper framework for offering and regulating AE in the UK. In particular, participants were asked to discuss their own choices and actions within their workplace pension and understanding of pensions’ information. Employers and professional advisers were also interviewed to build a complete picture of the practical implementation of AE. In considering these findings my work considers whether there are negative or unintended consequences of the policy change and how these are dealt with.

This year (2018), it is 25 years since the Momentum Strategy was first scientifically described. Despite this, the cause of the effect has not surely been concluded although it has been empirically studied in several previous studies. It has been shown to be valid for different kinds of assets. Since the authors of this thesis are based in Stockholm they thought it would be interesting and relevant to study if the strategy is valid on the Swedish market. The stock data comes from the stocks which has been part of the OMXS30 at least once during the period of 2010-2018. This study has also utilised two different ways on how to quantitate the return of the different portfolios. The effect of the holding period has in this report been attempted to address. The holding period is the length of the period which assets should be enclosed in the portfolio. One of the quantitation methods compared the portfolios’ development each month. The other method was more like a window analysis, to evaluate a portfolio’s return if one decides to invest in that theory until all the invested funds has been turned over. The study finds that the Momentum Strategy with holding periods of 2, 3 and 4 months significantly outperforms the market. With a higher significance level (10%) Momentum Strategy portfolios with holding periods of 2-6 and 11-14 months are outperforming the market. With a larger sample size, it is possible that the results would have been more conclusive.