The thesis compares several non-linear regression methods on synthetic data sets gen- erated using standard benchmarks for a continuous black-box optimization. For that com- parison, we have chosen the following regression methods: radial basis function networks, Gaussian processes, support vector regression and random forests. We have also included polynomial regression which we use to explain the basic principles of regression. The com- parison of these methods is discussed in the context of black-box optimization problems where the selected methods can be applied as surrogate models. The methods are evalu- ated based on their mean-squared error and on the Kendall's rank correlation coefficient between the ordering of function values according to the model and according to the function used to generate the data. 1

This master´s thesis evaluates the performance of the LAC, s.r.o. company. The theoretical part of the work introduces both a system of performance measurement and also the key principles of strategic analysis. Following this there is an introduction to the company, and to its business activities, together with an evaluation of its current status, utilising strategic and financial analysis. Based on the outputs of the analyses that were carried-out, the analytical part of this thesis presents the design of the Balanced Scorecard method. The final part of the work discusses a proposal for the implementation of a Balanced Scorecard in a corporate environment, with the aim of increasing the company performance and facilitating its future development.

本研究之主要目標在於建構一個能評估分散式關聯資料庫中之特色的需求導向績效評估方法。在過去的績效評估研究中，已經有許多人對於關聯式資料庫績效評估做了多方面的努力。但是，過去的關聯式資料庫資效評估方法如：Wisconsin、AS3AP、TPC系列的Benchmarks都有著一些限制及不足的地方。 過去的關聯式資料庫績效評估方法並無法完全的評估出分散式資料庫的特殊需求及其表現。所以本研究嘗試要建立出一個能專門適用於分散式資料庫導向的績效評估方法。為了要作出此績效評估方法，本研究採用了工作量模式的研究方法。先建出分散式資料庫績效評估的工作量模式，再以其來實作出績效評估方法。工作量模式分成三部分：資料模式、交易模式、控制模式。 / This thesis is intended to design a requirements-centric database benchmark, which can evaluate the general performance of the distributed relational database systems. In the past, there are many relational database benchmarks. But the relational database benchmarks like Wisconsin, AS3AP, TPC, TP1 have some constraints. In this study, we aim to design a general-purpose distributed database workload model and implement it. To design this benchmark, we need to build our workload model. The workload model consists of three components：data model, transaction model, control model. Each model has the requirement specification language to accommodate user's workloads.

Ce travail s'inscrit principalement dans le domaine de l'observation et de la commande non linéaires pour la machine asynchrone sans capteurs mécaniques. La première partie est consacrée aux problèmes d'observabilité auxquels la machine asynchrone est confrontée. Nous avons établi que La machine asynchrone est inobservable lorsqu'elle fonctionne à très basse vitesse sans capteurs mécaniques. Dans cet objectif, deux benchmarks sont définis : "Benchmark Observateurs sans capteurs mécaniques" et "Benchmark Commande sans capteurs mécaniques". Ces deux benchmarks permettent respectivement de tester les observateurs et les commandes de la machine asynchrone sans capteurs mécaniques sur des trajectoires difficiles liées notamment aux problèmes d'observabilité de cet actionneur à très basses vitesse. La deuxième partie est dédiée à la synthèse d'observateurs non linéaires (capteurs logiciels) pour la machine asynchrone qui permettent de remplacer l'information donnée par les capteurs physiques en utilisant comme seules mesures les grandeurs électriques. La dernière partie est consacrée à la conception d'une loi de commande non linéaire pour la machine asynchrone. Cette loi de commande est associée à un observateur de vitesse pour réaliser une commande sans capteurs mécaniques de la machine asynchrone. Les différents observateurs ainsi que la loi de commande ont été testés et validés expérimentalement au moyen des deux benchmarks sur la plate forme d'essai de machines électriques située à l'IRCCyN.

Conselho Nacional de Desenvolvimento CientÃfico e TecnolÃgico / This article analyzes the effect of decisions on resource allocation among items of operating expenses, administrative and otherwise, in the unilateral moments of the distribution of returns on Brazilian stock mutual funds. We follow methodologically Chambers, Chung and Fare (1996, 1998), using the Directional Distance Function (DDF) technique, which allows us to measure the behavior of the average value, and semicurtose semivariance of returns of funds, together with inputs associated with the internal allocation of resources. The boundary of "production" was constructed from the observation of the accounting entries monthly and daily returns for a panel of 59 mutual funds within the class of Ibovespa Activity during 2005 to 2009. We are able to evidence a high level of persistence in terms of efficiency, corroborating the evidence reported in Castro and Matos (2012). The more efficient funds have higher levels of cumulative return, except in 2008, lower levels of undesirable products associated with the risk and composition of operating expenses, such that the value of the ratio of non administrative expenses/ administrative expenses is approximately half of the observed for the most inefficient funds. / Este trabalho estuda os nÃveis de eficiÃncia do processo produtivo dos fundos de investimento em aÃÃes no Brasil, visando acomodar as crÃticas de Murthi et al. (1997) e Basso e Funari (2001). Analisa-se o efeito das decisÃes em termos de alocaÃÃo de recursos dentre rubricas de despesas especificamente operacionais ou administrativas nos nÃveis de ganho e de risco mensurado pela semivariÃncia e semicurtose, a partir da FunÃÃo DistÃncia Direcional proposta por Chambers, Chung e FÃre (1996, 1998). A observaÃÃo do comportamento conjunto dos inputs e outputs e a construÃÃo da fronteira de produÃÃo para um painel de 59 fundos mÃtuos de investimento em aÃÃo, categoria Ibovespa Ativo, durante 2005 a 2009, permitem evidenciar um elevado nÃvel de persistÃncia em termos de eficiÃncia, corroborando a evidÃncia de persistÃncia de performance reportada em Matos e Castro (2012). Os fundos sistematicamente eficientes sÃo todos administrados por instituiÃÃes financeiras privadas, possuem uma tendÃncia de alocar proporcionalmente mais em cotas de outros fundos que os fundos ineficientes e atendem essencialmente clientes institucionais, private ou de alta renda, enquanto os fundos ineficientes sÃo direcionados essencialmente para o pÃblico em geral. A eficiÃncia està associada ainda a maiores nÃveis de retorno acumulado, exceto em 2008, menores nÃveis de produtos indesejÃveis associados ao risco e a uma composiÃÃo dos gastos tal que, o valor da razÃo de gastos nÃo administrativos/gastos administrativos à aproximadamente a metade do evidenciado para os fundos sistematicamente ineficientes. Os resultados parecem robustos quando comparados aos tradicionais rankings de performance risco-retorno. Evidencia-se uma maior heterogeneidade de eficiÃncia no turbulento ano de 2008, caracterizada por uma maior quantidade de fundos na fronteira, quase 30%, e pelos maiores nÃveis de ineficiÃncia.

High Performance Computing (HPC) applications are data-intensive scientific software requiring significant CPU and data storage capabilities. Researchers have examined the performance of Amazon Elastic Compute Cloud (EC2) environment across several HPC benchmarks; however, an extensive HPC benchmark study and a comparison between Amazon EC2 and Windows Azure (Microsoft’s cloud computing platform), with metrics such as memory bandwidth, Input/Output (I/O) performance, and communication computational performance, are largely absent. The purpose of this study is to perform an exhaustive HPC benchmark comparison on EC2 and Windows Azure platforms. We implement existing benchmarks to evaluate and analyze performance of two public clouds spanning both IaaS and PaaS types. We use Amazon EC2 and Windows Azure as platforms for hosting HPC benchmarks with variations such as instance types, number of nodes, hardware and software. This is accomplished by running benchmarks including STREAM, IOR and NPB benchmarks on these platforms on varied number of nodes for small and medium instance types. These benchmarks measure the memory bandwidth, I/O performance, communication and computational performance. Benchmarking cloud platforms provides useful objective measures of their worthiness for HPC applications in addition to assessing their consistency and predictability in supporting them.

This thesis explores the potential use and implementation of a revenuemanagement model for coworking operators. With a critical realism philosophyand abductive approach, a quantitative study using primary data from a coworkingoperator has been conducted. Based on a comprehensive literature review, wehave found that much research is done on revenue management in the airline andhotel industries. However, we found no research on revenue management modelsthat intend to optimize revenue for coworking operations. Hence, this thesis aimsto fill this gap in existing academic research. Features from revenue managementmodels used in the hotel and airline industries are identified and analyzed with theobjective of implementing these in the coworking industry to efficiently maximizerevenue. The paper proposes the use of multinomial logit (MNL) model in theprocess of market segmentation; this method allows one to determine whichfactors influence the different segments. Moreover, the MNL model is used todefine the demand function from which a probability (probable?) distributionof total demand can be separated into demands representing each product class.Furthermore, the demand is used to calculate protection limits according to theExpected Marginal Seat Revenue (EMSR) model, with the objective of allocatingcapacity to the highest-yielding customers.Results indicate that the MNL regression is an effective tool to analyze themarket segmentation and demand allocation for coworking operators. Afterour successful analysis, we are prepared to argue with confidence that revenuemanagement models are applicable to coworking operations. / I denna uppsats undersöker vi den potentiella användningen och genomförandetav intäktsoptimeringsmodeller för coworking operatörer. En kvantitativ studiemed primär data från en coworking operatör har genomförts med en kritiskrealismfilosofi och ett abduktivt tillvägagångssätt. Det finns mycket forskningkring intäktsoptimering, framför allt inom flyg- och hotellbranschen, men ingensom behandlar intäktsoptimeringsmodeller med avseendepå coworking operatörer. Denna masteruppsats syftar till att bidra med kunskapför att fylla gapet kring revenue management för coworkingkontor, som saknas ibefintlig, svensk akademisk forskning idag.Vitala funktioner som utgör intäktoptimeringsmodeller ämnade för hotellochflygindustrin har identifierats och analyserats med målet att utforskamöjlig implementering för coworkingoperatörer. I uppsatsen genomförsen marknadssegmentering med hjälp av en multinomial regressionsanalys.Vidare görs en multinomial regressionsanalys med samtliga produktklasser somberoende variabler, för att få ut sannolikhetsfördelningen för vilka produkter somefterfrågas av den totala efterfrågan. Resultatet kan användas för att optimeratotala intäkterna genom att beräkna hur många platser som bör reserveras åthögt avkastande kunder, och hur många som kan hyras ut i tidigt skede. Förändamålet har vi tillämpat den så kallade Expected Marginal Seat Revenuemetoden, EMSR.Resultatet indikerar att multinomial logistisk regression är ett effektivt sätt attanalysera marknadssegment och styra efterfrågan till önskad produktklass. Samtatt användandet av rekommenderad revenue management modell är applicerbarpå coworking verksamheter. Alternativt: Resultatet indikerar att: i. multinomiallogistisk regression är ett effektivt sätt att analysera marknadssegment och styraefterfrågan till önskad produktklass. ii. användandet av rekommenderad revenuemanagement modell är applicerbar på coworking verksamheter.

In March 2018 the European Commission published the Action Plan on Financing Sustainable Growth. One of the main objectives with the actions presented in the action plan is to reorient capital flows towards sustainable investments, i.e. to influence more investors to invest sustainably. The action plan was followed by three proposals for transparency regulation regarding an EU taxonomy on sustainability, sustainability benchmarks and sustainability disclosures. Furthermore, the action plan included actions regarding two other transparency measures – sustainability labels and sustainability ratings. The first purpose of the thesis is to investigate if transparency regulation in the EU can be used as a tool to influence investors to choose sustainable investment funds. One of the main aims of the actions presented in the Action Plan on Financing Sustainable Growth, as well as the accompanying regulation proposals, is to reorient capital flows towards sustainable investments, i.e. to influence more investors to invest sustainably. In light of this, the Commission’s three proposed transparency regulations, as well as the concept of sustainability labels and ratings, are used as a basis for the investigation. The second purpose of the thesis is therefore to critically review the three regulation proposals and the concept of sustainability labels and ratings in order to gain an understanding of how different transparency measures can influence investors to choose sustainable investment funds. The transparency regulations and measures are analysed and critically reviewed in light of their objective to influence more investors to invest sustainably. A behavioural economics perspective, as well as consumer behaviour theories and decision-making models, are applied in order to analyse the transparency regulations and measures from an external perspective. Based on the analysis there are many indicators that transparency regulation can be used as a tool to influence investors to choose sustainable investment funds. However, to what extent transparency regulation can influence investor behaviour varies depending on which transparency measures are used and how they are designed. Sustainability benchmarks seem to have the least potential to influence investor behaviour, while the EU taxonomy on sustainability and sustainability labels seem to have the best potential to influence investor behaviour.

Analyzing the structure of a dataset is a challenging problem in high-dimensions as the volume of the space increases at an exponential rate and typically, data becomes sparse in this high-dimensional space. This poses a significant challenge to machine learning methods which rely on exploiting structures underlying data to make meaningful inferences. This dissertation proposes the <i>n</i>-TARP method as a building block for high-dimensional data analysis, in both supervised and unsupervised scenarios.<div><br></div><div>The basic element, <i>n</i>-TARP, consists of a random projection framework to transform high-dimensional data to one-dimensional data in a manner that yields point separations in the projected space. The point separation can be tuned to reflect classes in supervised scenarios and clusters in unsupervised scenarios. The <i>n</i>-TARP method finds linear separations in high-dimensional data. This basic unit can be used repeatedly to find a variety of structures. It can be arranged in a hierarchical structure like a tree, which increases the model complexity, flexibility and discriminating power. Feature space extensions combined with <i>n</i>-TARP can also be used to investigate non-linear separations in high-dimensional data.<br></div><div><br></div><div>The application of <i>n</i>-TARP to both supervised and unsupervised problems is investigated in this dissertation. In the supervised scenario, a sequence of <i>n</i>-TARP based classifiers with increasing complexity is considered. The point separations are measured by classification metrics like accuracy, Gini impurity or entropy. The performance of these classifiers on image classification tasks is studied. This study provides an interesting insight into the working of classification methods. The sequence of <i>n</i>-TARP classifiers yields benchmark curves that put in context the accuracy and complexity of other classification methods for a given dataset. The benchmark curves are parameterized by classification error and computational cost to define a benchmarking plane. This framework splits this plane into regions of "positive-gain" and "negative-gain" which provide context for the performance and effectiveness of other classification methods. The asymptotes of benchmark curves are shown to be optimal (i.e. at Bayes Error) in some cases (Theorem 2.5.2).<br></div><div><br></div><div>In the unsupervised scenario, the <i>n</i>-TARP method highlights the existence of many different clustering structures in a dataset. However, not all structures present are statistically meaningful. This issue is amplified when the dataset is small, as random events may yield sample sets that exhibit separations that are not present in the distribution of the data. Thus, statistical validation is an important step in data analysis, especially in high-dimensions. However, in order to statistically validate results, often an exponentially increasing number of data samples are required as the dimensions increase. The proposed <i>n</i>-TARP method circumvents this challenge by evaluating statistical significance in the one-dimensional space of data projections. The <i>n</i>-TARP framework also results in several different statistically valid instances of point separation into clusters, as opposed to a unique "best" separation, which leads to a distribution of clusters induced by the random projection process.<br></div><div><br></div><div>The distributions of clusters resulting from <i>n</i>-TARP are studied. This dissertation focuses on small sample high-dimensional problems. A large number of distinct clusters are found, which are statistically validated. The distribution of clusters is studied as the dimensionality of the problem evolves through the extension of the feature space using monomial terms of increasing degree in the original features, which corresponds to investigating non-linear point separations in the projection space.<br></div><div><br></div><div>A statistical framework is introduced to detect patterns of dependence between the clusters formed with the features (predictors) and a chosen outcome (response) in the data that is not used by the clustering method. This framework is designed to detect the existence of a relationship between the predictors and response. This framework can also serve as an alternative cluster validation tool.<br></div><div><br></div><div>The concepts and methods developed in this dissertation are applied to a real world data analysis problem in Engineering Education. Specifically, engineering students' Habits of Mind are analyzed. The data at hand is qualitative, in the form of text, equations and figures. To use the <i>n</i>-TARP based analysis method, the source data must be transformed into quantitative data (vectors). This is done by modeling it as a random process based on the theoretical framework defined by a rubric. Since the number of students is small, this problem falls into the small sample high-dimensions scenario. The <i>n</i>-TARP clustering method is used to find groups within this data in a statistically valid manner. The resulting clusters are analyzed in the context of education to determine what is represented by the identified clusters. The dependence of student performance indicators like the course grade on the clusters formed with <i>n</i>-TARP are studied in the pattern dependence framework, and the observed effect is statistically validated. The data obtained suggests the presence of a large variety of different patterns of Habits of Mind among students, many of which are associated with significant grade differences. In particular, the course grade is found to be dependent on at least two Habits of Mind: "computation and estimation" and "values and attitudes."<br></div>

Thesis (MA (Industrial Psychology))--University of Stellenbosch, 2008. / The survival of an organisation in a global competitive business environment depends on its survival skills (e.g. adaptability, flexibility and innovativeness), which is situated within its human resources. Hence, the recruitment function’s role is critical, since it is responsible for the procurement of key intellectual capital that could equip the organisation with a competitive advantage. However, against the accelerated growth of the Internet combined with global competition, traditional recruitment methods are increasingly failing to rise to the challenge of securing key intellectual capital, due to being too time consuming and expensive. Consequently, organisations are progressively turning to online recruitment (which does not refer to one specific technique, but rather several different Internet tools that can be employed as a recruitment method, such as organisational websites, specialised job sites, media websites or newsgroups). However, the presence of an organisational website that advertises vacancies is not sufficient to guarantee its success as a recruitment method. Hence, it is vital to gain an understanding of website recruitment, as well as how to maximise the website’s effectiveness and ability to facilitate successful online applicant behaviour (SOAB), in order to harness its full potential as a recruitment method, able to secure key intellectual capital. For the purpose of this study, SOAB refers to potential applicants (i.e. job seekers entering an organisational website in search of employment opportunities) being able to browse the website without any difficulty and obtain relevant and sufficient information concerning the organisation and its listed vacancies. In addition to enabling potential applicants to assess whether he/she will be able to fit, perform and accelerate in the work climate and culture of the organisation (i.e. decide whether he/she would like to work for the organisation), a website that facilitates SOAB, should also enable potential applicants to contact the organisation to address additional information needs. In essence, this study comprises of three phases and centres on the identification of website content- and usability design benchmarks that should contribute to a website’s ability to facilitate SOAB. By means of a comprehensive literature review, it is argued that the content- and usability design are vital contributing factors to a website’s ability to facilitate SOAB. It is also proposed that the effectiveness of the website as a recruitment method is, to a certain degree, linked to the effectiveness of the alignment of an organisation’s online recruitment strategy with the five stages of potential applicants’ job decision-making process (i.e. the recognition of an employment need, search for career related information, evaluation of career alternatives, identification and acceptance of employment and post-choice evaluation). Phase one, entails the identification of website content- and usability design benchmarks and culminates with the development of the Website Benchmarks Checklist. Phase two, which constitutes the overall purpose of this study, entails an audit of the sample of the ‘best’ SA employers’ (drawn from a survey conducted by the Corporate Research Foundation, 2005) websites. The primary aim is to determine the extent to which the sample’s website design incorporates the identified benchmarks. However, the quality of potential applicants’ interaction with a website (i.e. informative content being communicated in an effective, efficient and satisfactory manner), greatly influences their perception of the organisation’s image, its attractiveness as an employer, as well as their intention to pursue employment within the organisation. Hence, the third phase of the study entails the subjective evaluation of three websites (selected from the audited sample), by a sample of potential applicants. Throughout the study, the empirical tests conducted were descriptive in nature and utilised survey research methods to acquire the required data, related to the specified goals and objectives that encapsulate the aim and purpose of this study. The results obtained provided valuable insight into website design benchmarks that should assist potential applicants in their job decision-making process, increase a website’s ability to facilitate SOAB and maximise its effectiveness as a recruitment method responsible for securing key intellectual capital. In addition to revealing that the majority of the sample employs its website as a recruitment method, the audit results also indicated that although a high level of adherence existed concerning the sample’s incorporation of the usability design benchmarks. However, a notable difference that ranged from very little to relatively high existed with regards to the extent to which the sample’s website design adhered to the content design benchmarks. The subjective evaluation of the three websites by potential applicants revealed that in addition to being critical contributors to the quality of their interaction with a website, the content- and usability design also had a profound impact on their assessment of the websites. Finally, the results also showed that a similarity existed between potential applicants’ subjective evaluation and the extent to which the design of the three websites adhered to the recommended benchmarks (audit results).