Abstract : The objective is to provide methods to improve the performance, or prediction accuracy of standard stacking approach, which is an ensemble method composed of simple, heterogeneous base models, through the integration of the diversity generation, combination and/or selection stages for regression problems. In Chapter 1, we propose to combine a set of level-1 learners into a level-2 learner, or ensemble. We also propose to inject a diversity generation mechanism into the initial cross-validation partition, from which new cross-validation partitions are generated, and sub-sequent ensembles are trained. Then, we propose an algorithm to select best partition, or corresponding ensemble. In Chapter 2, we formulate the partition selection as a Pareto-based multi-criteria optimization problem, as well as an algorithm to make the partition selection iterative with the aim to improve more the ensemble prediction accuracy. In Chapter 3, we propose to generate multiple populations or partitions by injecting a diversity mechanism to the original dataset. Then, an algorithm is proposed to select the best partition among all partitions generated by the multiple populations. All methods designed and implemented in this thesis get encouraging, and favorably results across different dataset against both state-of-the-art models, and ensembles for regression. / Résumé : L’objectif est de fournir des techniques permettant d’améliorer la performance de l’algorithme de stacking, une méthode ensembliste composée de modèles de base simples et hétérogènes, à travers l’intégration de la génération de la diversité, la sélection et combinaison des modèles. Dans le chapitre 1, nous proposons de combiner différents sous-ensembles de modèles de base obtenus au primer niveau. Nous proposons un mécanisme pour injecter de la diversité dans la partition croisée initiale, à partir de laquelle de nouvelles partitions de validation croisée sont générées, et les ensembles correspondant sont formés. Ensuite, nous proposons un algorithme pour sélectionner la meilleure partition. Dans le chapitre 2, nous formulons la sélection de la partition comme un problème d’optimisation multi-objectif fondé sur un principe de Pareto, ainsi que d’un algorithme pour faire une application itérative de la sélection avec l’objectif d’améliorer d’avantage la précision d’ensemble. Dans le chapitre 3, nous proposons de générer plusieurs populations en injectant un mécanisme de diversité à l’ensemble de données original. Ensuite, un algorithme est proposé pour sélectionner la meilleur partition entre toutes les partitions produite par les multiples populations. Nous avons obtenu des résultats encourageants avec ces algorithmes lors de comparaisons avec des modèles reconnus sur plusieurs bases de données.

I quantified habitat selection for the endangered Zuni Bluehead Sucker Catostomus discobolus yarrowi and the Navajo Nation Genetic Subunit (NNGS) Bluehead Sucker Catostomus discobolus - a recent taxon described from genetic information. Both taxa are found in northern Arizona and New Mexico border regions. I examined fish [≥50 millimeters (mm) total length (TL)] selection of microhabitat conditions (i.e., water velocity, substrate size, overhead cover, water depth, instream cover, and mesohabitat conditions [i.e., pool, run riffle], during summer base flow conditions for NNGS Bluehead Suckers, and during both summer base flow and high spring flow conditions for Zuni Bluehead Suckers in six streams). Electrofishing, seining, and snorkeling were used to evaluate fish occupancy. From this information, I developed stream specific habitat suitability criteria (HSC) and then generalized HSC for each taxon, and tested transferability of the generalized HSC to individual streams. Zuni Bluehead Suckers and NNGS Bluehead Suckers occupied similar habitats: low velocity pools; sand, silt, and pebble substrate; high percent of instream cover; and water temperatures ranging from 2-21°C. However, Zuni Bluehead Suckers selected for low (0-25%) overhead cover where as NNGS Bluehead Sucker selected for high (0-75%) overhead cover. This was likely due to the source of instream cover–aquatic macrophytes that required sunlight in the Zuni Bluehead Sucker streams, and large woody debris falling from overhead branches in the NNGS Bluehead Sucker streams. Suggestions for managers includes maintaining existing cover or artificially construct additional instream cover; promote overhead cover (e.g., maintaining large trees along streams) and pool mesohabitats. In addition to this work I also tested the new method of environmental DNA (eDNA) to further help conservation efforts for these taxa. Environmental DNA has typically been used to detect invasive species in aquatic environments through water samples. I compared the efficacy of eDNA methodology to American Fisheries Society standard snorkeling surveys to detect presence of a rare fish species. My study site included three streams on the Navajo Nation in northern Arizona and northern New Mexico containing Navajo Nation Genetic Subunit Bluehead Sucker Catostomus discobolus and the Zuni Bluehead Sucker Catostomus discobolus yarrowi. To determine sample sites, I first divided entire wetted area of streams into 100-m consecutive reaches. I systematically selected 10 of those reaches for snorkel and eDNA surveys. Water samples were taken in 10-m sections within each 100-m reach, and fish presence via snorkeling was noted in each 10-m section as well. Water samples were collected at the downstream starting point of each reach, and continued upstream in each section 5 to 8 m ahead of the snorkeler. A qPCR was run on each individual water sample in quadruplicate to test for sucker presence or absence. I was able to positively detect both species with eDNA sampling techniques in two out of three streams. Snorkeling resulted in positive detections of both species in all three streams. In streams where fish were detected with eDNA sampling, snorkeling detected fishes at 11-29 sites per stream, where as eDNA detected fish at 3-12 sites per streams. My results suggested that AFS standard snorkeling was more effective at detecting target fish species than eDNA. To improve eDNA sampling, the amount of water collected and tested should be increased. Additionally, filtering water on site may improve eDNA techniques for detecting fish. Future research should focus on standardizing eDNA sampling to provide a widely operational sampling tool similar to electrofishing, netting, and hydroacoustics.

ITC/USA 2009 Conference Proceedings / The Forty-Fifth Annual International Telemetering Conference and Technical Exhibition / October 26-29, 2009 / Riviera Hotel & Convention Center, Las Vegas, Nevada / Virtualization technologies have the potential to transform the telemetry post-processing environment. Significant efficiencies can be gained by migrating telemetry post processing activities to virtual computing platforms. However, while facilitating better server utilization, virtualization also presents several challenges; one of the most difficult of those challenges being security. In virtualization, server environments are replicated in software; unfortunately, the security individual servers provide is not replicated in a software stack implementation of a server environment. After reviewing virtualization fundamentals, security issues and their impact on telemetry post processing will be discussed.

This thesis induces a theory for the pre-measurement phase of the asset recognition process in the financial reporting domain centred upon the use of the induced artefact-based asset recognition criteria which are applicable to all assets. In common with standard-setting bodies, such as the International Accounting Standards Board (IASB), I adopt a social constructionist stance (Miller, 1994). It is one that is constructed from a process of consultation. I consult in order to, first, explore the accounting asset recognition process from a conceptual point of view and, second, so that my social construction in that regard can be legitimised, for the most part, on the basis of a consensus of those consulted. However, unlike the standard setters‘ regulatory process, my analysis is structured using a grounded theory approach. The target audience comprises those experts who have been and/or who are currently involved in some way with the development of the IASB‘s conceptual framework (CF) project, including IASB board members. Different data collection methods were adopted combing both qualitative and quantitative data. In respect of the qualitative research, I carried out two sets of interviews. The first set was conducted with Canadian Accounting Standards Board members in May 2008 and International Accounting Standard Board members in June, 2008. The second set was conducted with more International Accounting Standard Board members, UK-Accounting Standard Baord members and other experts within the area being studied. These two sets of interviews were useful for determining the preliminary concepts and categories in the open coding and axial coding structure. In respect of the quantitative research, the concepts and categories raised from the first two sets of interviews were then used to construct an on line questionnaire. The questionnaires were emailed to national standard setters in Canada, the USA, Australia, Germany and the United Kingdom. This has been followed up with an interview with UK ASB director to help in finalising the theory saturation and to validate the reliability of the generated theory. The generated theory demonstrates a three-circled set of criteria for the pre-measurement phase of an asset recognition process. The three-circled set of asset recognition criteria presented in this thesis breaks free from the narrow definitional and rule based perspective of accounting epistemology to offer an alternative view based on the recognition of artefacts.

The primary purpose of this thesis is to investigate the operation of membership criteria, which are applied against countries that wish to join the European Union (EU). More specifically, the importance of some criteria in comparison to others is considered. To answer this proposal, three questions are posed: What are the EU membership criteria? Are some criteria more important than other criteria? Does the EU treat candidate countries differently in applying the membership criteria? Three countries are used in this analysis, each country representing one recent enlargement round: Poland (2004), Romania (2007), and current candidate country Turkey. Both official documents, such as EU treaties, and unofficial documents such as statements from EU leaders are used to analyze the application of membership criteria. The method that is used in this thesis is idea analysis. The EU has both official criteria, which are found in the foundation treaties, and unofficial criteria, which are the public and political opinions among the candidate countries and current EU member states. It is suggested that none of the criteria are more important than others; a candidate country must fulfill virtually all the official and unofficial criteria in order to gain EU membership. Although all criteria must be met, the application of the criteria is uneven between candidate countries. The EU does not treat countries differently when it comes to the official criteria. However, when it comes to the unofficial criteria, Turkey is treated differently from Romania and Poland. There is greater resistance to Turkey’s future membership, making it much more difficult for Turkey to reach the stage of full EU membership.

Denna studie fokuserar på att klassificera hat-meddelanden riktade mot mobiloperatörerna Verizon,  AT&amp;T and Sprint. Huvudsyftet är att med hjälp av maskininlärningsalgoritmen Support Vector Machines (SVM) klassificera meddelanden i fyra kategorier - Hat, Orsak, Explicit och Övrigt - för att kunna identifiera ett hat-meddelande och dess orsak. Studien resulterade i två metoder: en "naiv" metod (the Naive Method, NM) och en mer "avancerad" metod (the Partial Timeline Method, PTM). NM är en binär metod i den bemärkelsen att den ställer frågan: "Tillhör denna tweet klassen Hat?". PTM ställer samma fråga men till en begränsad mängd av tweets, dvs bara de som ligger inom ± 30 min från publiceringen av hat-tweeten. Sammanfattningsvis indikerade studiens resultat att PTM är noggrannare än NM. Dock tar den inte hänsyn till samtliga tweets på användarens tidslinje. Därför medför valet av metod en avvägning: PTM erbjuder en noggrannare klassificering och NM erbjuder en mer utförlig klassificering. / This study focused on finding the hate tweets posted by the customers of three mobileoperators Verizon, AT&amp;T and Sprint and identifying the reasons for their dissatisfaction. The timelines with a hate tweet were collected and studied for the presence of an explanation. A machine learning approach was employed using four categories: Hate, Reason, Explanatory and Other. The classication was conducted with one-versus-all approach using Support Vector Machines algorithm implemented in a LIBSVM tool. The study resulted in two methodologies: the Naive method (NM) and the Partial Time-line Method (PTM). The Naive Method relied only on the feature space consisting of the most representative words chosen with Akaike Information Criterion. PTM utilized the fact that the majority of the explanations were posted within a one-hour time window of the posting of a hate tweet. We found that the accuracy of PTM is higher than for NM. In addition, PTM saves time and memory by analysing fewer tweets. At the same time this implies a trade-off between relevance and completeness. / <p>Opponent: Kristina Wettainen</p>

This thesis is the result of applying a novel methodology which I labelled ‘participative foresight scenario mapping’. This methodology couples participatory methods for building holistic foresight scenarios for sustainable development in Flores Island (Azores, Portugal) with a multi-criteria appraisal method, Multi-criteria mapping (Stirling, 1997), to assess these scenarios alongside five sector based regional scenarios (Secretaria Regional do Ambiente e do Mar, 2006). The main research question was to reflect on how small isolated societies, which have a distant relation with strategic decision-making centres, can define their transitions to sustainability. Small islands represent interesting cases to reflect on sustainability, these small territories distant from main decision-making centres challenge decision-making and require a consideration of the issues of scale. Islands have also been seen as small, manageable models of the world, providing the opportunity to explore innovative solutions at a scale that allows inclusion of as many different factors as possible. Small islands’ populations are especially linked to their island and they develop, by the effects of isolation, a strong particular relation to the place, the role of identity is then crucial in fostering sustainable practices adapted to the island. A succession of individual scoping interviews with twenty four regional and local decision-makers and key informants and seven focus groups with a total of thirty local lay citizens gave me the opportunity to develop two differentiated multi-sector scenarios for Flores Island which were identified as Standard and Balanced development scenarios. The Balanced development scenario reflects a desire to develop an island that bases its economy on greater self-sufficiency for agricultural products, quality and certified products, and natural conservation and valorisation. The Standard development scenario is based on economic growth through tourism and primary sector intensification, and public investment in infrastructures; this scenario can be summarized as the continuation of the actual model of development. The appraisal of both holistic narratives allows in depth exploration of the complex issues related to sustainability, such as the preference between weak and strong sustainability, that otherwise would have been too difficult to assess by such a variety of research participants. Working with holistic scenarios raised the limits of the capacity to show proficiency in a wide variety of fields. The research demonstrated the feasibility of applying the multi-criteria mapping method to support the analysis of holistic non-technical scenarios. The combination of qualitative and quantitative data brought depth to the analysis and improved the understanding of the desired sustainable futures in islands. But the quantitative appraisal was overshadowed by strong uncertainties that made difficult the identification of a best scenario. Uncertainty was explained by the risks inherent to the scenarios, the limited expertise in all the criteria, the complexity of the holistic scenarios, the time horizon (20 years), doubts on the effective implementation of the chosen scenario, and the existence of potentially disrupting external factors. The process was also the opportunity to understand the role that social capital might play in the transition to the desired future for this island. It is shown in the thesis that a successful transition to sustainable development can only be reached if the objectives are understood and shared by the population.

The approach to water management worldwide is currently in transition, with a shift from centralised infrastructures to greater consideration of decentralised technologies, such as rainwater harvesting (RWH). Initiated by recognition of drivers, including water demand, increasing risk of ground-water pollution and flooding, the value of RWH is filtering across the academic-policy boundary. However, in Nigeria, implementation of sustainable water management (SWM), such as RWH systems, is inefficient social, environmental and technical barriers, concerns and knowledge gaps exist, which currently restrict its widespread utilisation. This inefficiency contributes to water scarcity, water-borne diseases, and loss of lives and property due to flooding. Meanwhile, several RWH technologies have been developed to improve SWM through both demand and storm-water management. Such technologies involve the use of storage tanks, surface water reservoirs and ground-water recharge pits as storage systems. A framework was developed to assess the significance and extent of water management problems, match the problems with existing RWH-based solutions and develop a robust ready-to-use multi-criteria analysis tool that can quantify the costs and benefits of implementing several RWH-based storage systems. The methodology adopted was the mixed method approach, involving a detailed literature review, followed by a questionnaire survey of 1067 household respondents, 135 Nigerian Architects and Civil Engineers and focus group discussion with Stakeholders. A total of 1042 sets of data were collected through a questionnaire survey and analysed using SPSS, Excel and selected statistical methods to derive weightings of the attributes for the tool. Following this, three case studies were selected to collect data for hydrological modelling using the RainCycle model. From the results it is found that the most important barrier constraining sustainable RWH regime in Ibadan was obsolete and insufficient operational equipment, followed by poor renumeration of water corporation staff and misuse of available funds. In addition, the measure of importance of storage capacity was established, with the highest score of 4.5 which reflects the general inadequacy of storage as a major barrier to the adoption of RWH as a sustainable water management method. Further, respondents’ major health hazards associated with drinking contaminated water was established. A larger proportion (61.2%) of respondents chose prevalence of typhoid fever; some have a prevalence of diarrhea (19.4%), while few of respondents’ water sources is free from water-borne diseases (2.3%). The tool developed is an integrated platform of related evaluation techniques, including Whole Life Cycle Cost Analysis and Multi-Attribute Utility Theory. The tool uses data including cost and quantities of materials for building a RWH storage system and quantifies the cost and benefits of alternative RWH-based systems that can improve project management. This tool is novel, given its integration of the analytical techniques mentioned above and application for selecting the most appropriate RWH-based SWM systems. The implementation of the tool is envisaged to provide an objective platform for the quantification of the costs and benefits of RWH-based systems prior to implementation.

Kuwait’s per capita electrical energy consumption is among the largest in the world, reaching 13,663 kWh per person in 2011. The electricity demand in Kuwait is increasing, which requires additional investments in power generation. A particular challenge in Kuwait is the peak demand in summer, when extreme heat increases air conditioning loads. Peak demand reached 11,220 MW in 2011, with a fast growth rate averaging 5.6% over the last decade and a maximum production capacity of around 14,720 MW. It is not possible to cope with this demand simply by increasing generation capacity. Therefore, the only alternative available to Kuwait is to control electricity demand via demand-side management. The main objective of this research is to assess and select the optimal demand- side management (DSM) technologies for buildings in the governmental sector (office, religious and school buildings) and to investigate the key factors in the evaluation process. To achieve the research objective, a model was proposed using multi-criteria decision-making techniques to enable the forecasting and comparison of DSM alternatives that are suitable for buildings. The developed model includes environmental, economic, technical and social considerations. The research methodology is based on three main phases. Phase 1: Data collection through a mail questionnaire that was sent to 42 experts for the identification of criteria and demand-side management alternatives. Phase 2: Screening and narrowing of the data collected in Phase 1 using a set of questionnaires, including the identification of potential DSM alternatives and criteria suitable for further analysis. This phase was performed through the Delphi process, taking into consideration the opinions of 28 experts. Phase 3: Use of Multi-criteria Decision Making (MCDM) techniques, namely the Analytic Hierarchy Process (AHP) and Fuzzy Analytic Hierarchy Process (FAHP), to evaluate and rank the identified DSM alternatives and criteria. This was done using pairwise comparisons of 17 experts to evaluate the criteria and alternatives for the buildings (office, religious and school). The research showed that the experts identified six alternative technologies and five criteria for the selection and evaluation of governmental buildings (office, religious and school). After performing the three phases of this research project, a set of criteria and alternatives were ranked based on the information gathered from every group of experts. It was found that reduction in consumption, capital cost and ease of implementation were the three most recommended criteria for the selection of DSM technologies in Kuwait government buildings while high efficiency lighting and programmable thermostats were identified as the most recommended DSM technologies for these buildings. An important aspect of this research is that unlike engineering approaches which sometimes depend on expensive test equipment or, often for building design, expensive computer modelling exercises, the proposed framework can be easily adopted by anyone without any significant financial cost. The FAHP approach was also tested and its results compared with those of AHP. There was a slight difference between using AHP and FAHP in terms of ranking the criteria and alternatives but that the difference barely affects the ranking. It was found that the AHP provides a convenient MCDM approach for solving the complex problem of selecting the optimal DSM options for buildings. The contributions of this thesis are the development of a novel framework for systematic selection and ranking of DSM technologies in different types of buildings using the Delphi method and AHP; the identification of the most important DSM technologies and criteria for their selection for three types of governmental buildings in Kuwait; and establishing that results from using AHP and FAHP for the selection of appropriate DSM measures in these buildings are almost identical, so use of AHP is likely to be sufficient in most building studies of this type.

This thesis deals with control of stock in an inventory, focusing on inventory placement. The purpose of this thesis is to reduce the transport distance within the main stock house while gathering inventory. This will be achieved by reconstructing the inventory placement in consideration with how frequently the inventories get picked and mass of the inventory. In particular, the literature and the data that is collected from the company´s business system have laid the foundation for the thesis. In general, interviews and observations also contributed to the data collection. To fulfill the aim and to produce arbitrary results, two issues have been developed regarding which attributes that should determine the position of the inventory in the stock house and how to obtain a more effective inventory structure? The authors have jointly produced a result of suggestions for future inventory placement in terms of picking frequency and weight. Initially a situation analysis was conducted to identify known problems with the inventory´s placement and storage systems. The problems that were identified were that the inventory placement has no consideration regarding picking frequency. To determine the most frequent picked inventory an ABC analysis was conducted. All of the inventories were spread out throughout the whole stock house. To take in account, the additional criterion, which was weight, a multi-criteria analysis was performed in combination with the ABC analysis. The results of the combined analysis provided that the basis for drawing up concepts for future inventory placement. The proposal includes optimized inventory placements in different zones of the most frequently picked inventory with weight as an additional criterion. / Arbetet behandlar området lagerstyrning med inriktning på optimering av artikelplacering. Syftet med arbetet är att minska transportsträckor inom centrallagret, detta genom en omstrukturerad artikelplacering med hänsyn till plockfrekvens och tyngd. I synnerhet har litteratur samt insamlad data från företagets affärssystem lagt grunden för arbetet. I allmänhet har intervjuer och observationer också bidragit till datainsamling. För att uppfylla arbetets syfte och ta fram ett godtyckligt resultat har två frågeställningar tagits fram angående vilka attribut som bör avgöra en artikels position i lagret samt hur en effektivare lagerstruktur kan erhållas. Författarna har gemensamt arbetat fram ett resultat i form av ett förslag för framtida artikelplacering sett till plockfrekvens och tyngd. Inledningsvis genomfördes en nulägesanalys för att identifiera kända problem med artikelplacering och lagersystem. De problem som identifierades var att artikelplaceringen sker utan underlag för plockfrekvens. För att utröna de mest frekventa artiklarna genomfördes en ABC-analys. Samtliga artiklar lokaliserades utspridda över hela lagret. För att ta hänsyn till ytterligare ett kriterium, vilket var tyngd, genomfördes multikriterieanalys i kombination med ABC-analys. Resultatet av den kombinerade analysen gav underlag för att ta fram förslag för framtida artikelplacering. Förslaget innehåller optimerade artikelplaceringar inom olika zoner för de mest plockfrekventa artiklarna med tyngd som ytterligare kriterium.