Value of information is one of the key features of decision analysis. This work deals with providing a consistent and functional methodology to determine VOI on proposed well tests in the presence of uncertainties. This method strives to show that VOI analysis with the help of discretized versions of continuous probability distributions with conventional decision trees can be very accurate if the optimal method of discrete approximation is chosen rather than opting for methods such as Monte Carlo simulation to determine the VOI. This need not necessarily mean loss of accuracy at the cost of simplifying probability calculations. Both the prior and posterior probability distributions are assumed to be continuous and are discretized to find the VOI. This results in two steps of discretizations in the decision tree. Another interesting feature is that there lies a level of decision making between the two discrete approximations in the decision tree. This sets it apart from conventional discretized models since the accuracy in this case does not follow the rules and conventions that normal discrete models follow because of the decision between the two discrete approximations. The initial part of the work deals with varying the number of points chosen in the discrete model to test their accuracy against different correlation coefficients between the information and the actual values. The latter part deals more with comparing different methods of existing discretization methods and establishing conditions under which each is optimal. The problem is comprehensively dealt with in the cases of both a risk neutral and a risk averse decision maker. / text

We have analyzed the application of decision analysis and risk management tools to solve practical problems associated with Climate Change and Risk Detection in the financial services industry. Geoengineering, which is described as an intentional modification of earth’s environment to mitigate the harmful effects of climate change, is evaluated as a policy alternative using the aforementioned tools. We compared the performance of geoengineering with optimal emission controls and a business as usual strategy under various scenarios and found that geoengineering passes the cost benefit test for a majority of the scenarios. We modified the DICE model (Nordhaus, 2008) and used it to evaluate the performance of different environmental policies. Our results show geoengineering as a potential alternative to solve climate change problems. Through this application, and by comparing our findings against Goes et al. (2011), we showed that how framing of the decision problem can lead to completely different results. We also analyzed the application of risk management in the financial services industry. The industry faces three main types of risk: Market risk, Credit risk and Operational risk. Market risk is managed using a diversified portfolio, derivatives, insurance and contracts. More challenging is the task of preventing credit and fraud risk. Statistical models used by the industry to detect and prevent these types of risk are explained in the thesis. / text

Stochastic price models have proven material to decision making in the oil industry when accurate valuations are important, but little consideration is given to their impact on decisions based on relative project rankings. Traditional industry economic analysis methods do not usually consider uncertainty in oil price, although the real options literature has shown that this practice underestimates the value of projects that have flexibility. Monetary budget constraints are not always the limiting constraints in decision making; there may be other constraints that limit the number of projects a company can undertake. We consider building a portfolio of upstream petroleum development projects to determine the relevance of stochastic price models to a decision for which accurate valuations may not be important. The results provide guidelines to determine when a stochastic price model should be used in economic analysis of petroleum projects. / text

Transformations of the various agricultural systems have been taking place in the coastal zone of Bangladesh. While some farmers continue to follow traditional practices, in recent years, others have become involved in massive shrimp cultivation, shrimp-rice cultivation, a rice-based improved agricultural system or a shrimp-rice-vegetable integrated system. All these types of agriculture are being practiced under highly vulnerable environmental conditions. The long-term livelihood, food security and adaptation of the coastal people largely depend on the sustainability of these agricultural practices. In this context, assessing the level of sustainability is extremely important and will be essential for developing future policy options in Bangladesh. This study attempts to examine the sustainability of agricultural practices in the coastal region of Bangladesh. A field study was carried out in 2011 in five villages of five upazilas in the mature and active delta areas of the country. The data were collected through in-depth questionnaire surveys, focus groups discussions, field observation, key informants and secondary materials. A comprehensive suite of indicators was developed considering productivity, efficiency, stability, durability, compatibility and equity of the coastal agriculture. The categories and the indicators were weighted using Multi-Criteria Decision Analysis (MCDA) to measure the sustainability level of five study sites. The integrated agricultural system (shrimp-rice-vegetable) of Dumuria appeared to be the most sustainable system among agricultural practices, and other integrated systems (rice-based improved agricultural system) of Kalaroa were also found to show a good level of sustainability. The massive shrimp cultivation system of Shyamnagar and Kaliganj appears to be least sustainable. A traditional agriculture system with some improved methods followed in Bhola Sadar also performed in a satisfactory manner, but there were limitations in terms of its location in the active delta. The level of the sustainability measured in this study allows for a comparison among agricultural practices of the five study sites. The information generated from the study may be used in formulating policies for this part of the country. Measuring agricultural sustainability in this way produces a useful summary of sustainability issues and also provides some vital learning experiences. A holistic and interdisciplinary approach is attempted in this study for assessing and comparing the sustainability level of coastal agricultural systems. It has the potential to become useful as one of the frameworks for sustainability assessment. / Thesis (Master, Environmental Studies) -- Queen's University, 2012-09-28 15:08:18.847

In this dissertation, we focus on the problem of self-management in distributed systems. In this context, we propose a new methodology for reactive self-management based on multiple criteria decision analysis (MCDA). The general structure of the proposed methodology is extracted from the commonalities of the former well-established approaches that are applied in other problem domains. The main novelty of this work, however, lies in the usage of MCDA during the reaction processes in the context of the two problems that the proposed methodology is applied to. In order to provide a detailed analysis and assessment of this new approach, we have used the proposed methodology to design distributed autonomous agents that can provide self-management in two outstanding problems. These two problems also represent the two distinct ways in which the methodology can be applied to self-management problems. These two cases are: 1) independent self management, and 2) coordinated self-management. In the simulation case study regarding independent self-management, the methodology is used to design and implement a distributed resource consolidation manager for clouds, called IMPROMPTU. In IMPROMPTU, each autonomous agent is attached to a unique physical machine in the cloud, where it manages resource consolidation independently from the rest of the autonomous agents. On the other hand, the simulation case study regarding coordinated self-management focuses on the problem of adaptive routing in mobile ad hoc networks (MANET). The resulting system carries out adaptation through autonomous agents that are attached to each MANET node in a coordinated manner. In this context, each autonomous node agent expresses its opinion in the form of a decision regarding which routing algorithm should be used given the perceived conditions. The opinions are aggregated through coordination in order to produce a final decision that is to be shared by every node in the MANET. Although MCDA has been previously considered within the context of artificial intelligence---particularly with respect to algorithms and frameworks that represent different requirements for MCDA problems, to the best of our knowledge, this dissertation outlines a work where MCDA is applied for the first time in the domain of these two problems that are represented as simulation case studies. / Graduate

In several types of embedded systems the applications are deployed both as software and as hardware components. For such systems, the partitioning decision is highly important since the implementation in software or hardware heavily influences the system properties. In the industry, it is rather common practice to take deployment decisions in an early stage of the design phase and based on a limited number of aspects. Often such decisions are taken based on hardware and software designers‟ expertise and do not account the requirements of the entire system and the project and business development constraints. This approach leads to several disadvantages such as redesign, interruption, etc. In this scenario, we see the need of approaching the partitioning process from a multiple decision perspective. As a consequence, we start by presenting an analysis of the most important and popular Multiple Criteria Decision Analysis (MCDA) methods and tools. We also identify the key requirements on the partitioning process. Subsequently, we evaluate all of the MCDA methods and tools with respect to the key partitioning requirements. By using the key partitioning requirements the methods and tools that the best suits the partitioning are selected. Finally, we propose two MCDA-based partitioning processes and validate their feasibility thorough an industrial case study.

Municipalities increasingly have less financial resources to spend on implementation of transport strategies and plans. This situation is putting pressure on transport professionals to minimize wasteful expenditure on projects that do not deliver high transport service improvements. As such, the need for efficient, pragmatic decision making on policy direction, infrastructure expenditure, or any transport interventions is becoming very critical. Thus, transport professionals are increasingly in need of tools to help them predict with increased accuracy the outcomes of their intended transport interventions. The City of Cape Town has a Bus Rapid Transport system called MyCiTi. Current MyCiTi operations are incurring losses. The service is kept running on the back of subsidies from the federal government. There is a need for rationalization of the system. However, with strained resources, the interventions on the system have to guarantee improvements. Overemphasis on the ability of MyCiTi BRT service to support transportation during the 2010 soccer world cup event heavily influenced the design of the network. As a result, network appraisal is one area that can be done on the system to identify areas of improvement. In this thesis, decision making support will be demonstrated using a network design appraisal process for the MyCiTi BRT system in Cape Town. The existing MyCiTi network will undergo network improvement using heuristic node insertion technique leading to multiple network scenarios in a modeling environment. Agent-Based demand mobility behavior simulation will be used on each of the network scenarios to come up with network performance indicators. These network performance indicators will be used in the multi-criteria decision analysis (MCDA) model to come up with a ranking of the network scenarios and help in deciding on the optimum network improvement intervention. Overall, findings of this research show the importance of weighting of the performance indicators. Where networks that score well in the performance indicator with the high weights also rank high. In conclusion, the study has demonstrated the importance of decision making support in interventions on complex systems like bus systems. Recommendations on the possible avenues of research stemming from this thesis have also been outlined.

L’aide multicritère à la décision (AMCD) vise à faciliter et améliorer la qualité du processus de prise de décision. Les méthodes d’AMCD permettent de traiter les problèmes de choix, rangement et classification. Ces méthodes impliquent généralement la construction d’un modèle. Déterminer les valeurs des paramètres de ces modèles n’est pas aisé. Les méthodes d’apprentissage indirectes permettent de simplifier cette tâche en apprenant les paramètres du modèle de décision à partir de jugements émis par un décideur tels que “l’alternative a est préférée à l’alternative b” ou “l’alternative a doit être classifiée dans la meilleure catégorie”. Les informations données par le décideur sont généralement parcimonieuses. Le modèle d’AMCD est appris au cours d’un processus interactif entre le décideur et l’analyste. L’analyste aide le décideur à formuler et revoir ses jugements si nécessaire. Le processus s’arrête une fois qu’un modèle satisfaisant les préférences du décideur a été trouvé. Le “preference learning” (PL) est un sous domaine du “machine learning” qui s’intéresse à l’apprentissage des préférences. Les algorithmes de ce domaine sont capables de traiter de grands jeux de données et sont validés au moyen de jeux de données artificiels et réels. Les jeux de données traités en PL sont généralement collectés de différentes sources et sont entachés de bruit.Contrairement à l’AMCD, il existe peu ou pas d’interaction avec l’utilisateur en PL. Le jeu de données fourni en entrée à l’algorithme est considéré comme un échantillon éventuellement bruité d’une “réalité” ou “vérité de terrain”. Les algorithmes utilisés dans ce domaine ont des propriétés statistiques fortes leur permettant de s’affranchir du bruit dans ces jeux de données. Dans cette thèse, nous développons des algorithmes d’apprentissage permettant d’apprendre lesparamètres de modèles d’AMCD. Plus précisément, nous développons une métaheuristique afin d’apprendre les paramètres d’un modèle appelé MR-Sort (“majority rule sorting”). Cette métaheuristique est testée sur des jeux de donnéesartificiels et réels utilisés dans le domaine du PL. Nous utilisons cet algorithme afin de traiter un problème concret dans le domaine médical. Ensuite nous modifions la métaheuristique afin d’apprendre les paramètres d’un modèle plus expressif appelé NCS (“non-compensatory sorting”). Finalement, nous développons un nouveau type de règle de veto pour les modèles MR-Sort et NCS qui permet de prendre les coalitions de critères en compte. La dernière partie de la thèse introduit les méthodes d’optimisation semi-définie positive (SDP) dans le contexte de l’aide multicritère à la décision. Précisément, nous utilisons l’optimisation SDP afin d’apprendre les paramètres d’un modèle de fonction de valeur additive. / Multiple-criteria decision analysis (MCDA) aims at providing support in order to make a decision. MCDA methods allow to handle choice, ranking and sorting problems. These methods usually involve the elicitation of models. Eliciting the parameters of these models is not trivial. Indirect elicitation methods simplify this task by learning the parameters of the decision model from preference statements issued by the decision maker (DM) such as “alternative a is preferred to alternative b” or “alternative a should be classified in the best category”. The information provided by the decision maker are usually parsimonious. The MCDA model is learned through an interactive process between the DM and the decision analyst. The analyst helps the DM to modify and revise his/her statements if needed. The process ends once a model satisfying the preferences of the DM is found. Preference learning (PL) is a subfield of machine learning which focuses on the elicitation of preferences. Algorithms in this subfield are able to deal with large data sets and are validated withartificial and real data sets. Data sets used in PL are usually collected from different sources and aresubject to noise. Unlike in MCDA, there is little or no interaction with the user in PL. The input data set is considered as a noisy sample of a “ground truth”. Algorithms used in this field have strong statistical properties that allow them to filter noise in the data sets.In this thesis, we develop learning algorithms to infer the parameters of MCDA models. Precisely, we develop a metaheuristic designed for learning the parameters of a MCDA sorting model called majority rule sorting (MR-Sort) model. This metaheuristic is assessed with artificial and real data sets issued from the PL field. We use the algorithm to deal with a real application in the medical domain. Then we modify the metaheuristic to learn the parameters of a more expressive model called the non-compensatory sorting (NCS) model. After that, we develop a new type of veto rule for MR-Sort and NCS models which allows to take criteria coalitions into account. The last part of the thesis introduces semidefinite programming (SDP) in the context of multiple-criteria decision analysis. We use SDP to learn the parameters of an additive value function model.

The problem of project selection is of significant importance in management of information systems. Almost $2 trillion is spent worldwide every year on IT projects, with over $600 billion spent in the US alone. Traditionally, managers have being using the classical net present value (NPV) method in conjunction with multicriteria scoring models for ROI analysis and selection of IT project investments The multicriteria models use ad-hoc evaluation criteria to assign priority weights and then rate the alternatives against each criterion. These models have two limitations. First, the criteria and weights are based on subjective judgments, allowing the introduction of politics in the information management decision process and the generation of arbitrary results. Second, the classical approach uses deterministic estimations of the cost, benefits and the returns of the projects, without considering the impact of uncertainty and risk in the business decisions. This research proposed a better alternative for ROI analysis and selection of IT projects using a real option strategic scorecard (ROSS) approach. In contrast with traditional methodologies and previous research work, the ROSS decision framework uses a more comprehensive, axiomatic approach for systematically measuring both the business value and the strategic implications of IT project investments. The ROSS approach integrates in a unified IT project management decision framework the best elements of real option theory, strategic balanced scorecards, Monte Carlo simulations and analytical network processes to fully analyzes the effect of uncertainty and risk in the IT investment decisions. In addition, the ROSS approach complies with the critical success factors that have being identified in the literature for validation of IT decision frameworks. The main benefit of the ROSS approach is to enable managers to better compare and rank projects in the IT portfolio, optimizing the ROI analysis and selection of information system projects.

The thesis is divided into two parts concerning different topics. The first is solving a multi-period portfolio decision problem, and the second, more theoretical, is a numerical comparison of uncertainty measures within evidence theory. Nowadays, portfolio problems are very common and present in several fields of study. The problem is inspired by a real-world infrastructure manage- ment case in the energy distribution sector. The problem consists of the optimal selection of a set of activities and their scheduling over time. In scheduling, various constraints and limits that the company has to meet must be considered, and the selection must be based on prioritizing the activities with a higher priority value. The problem is addressed by Port- folio Decision Analysis: the priority value of activities is assigned using the Multi-Attribute Value Theory method, which is then integrated with a multi-period optimization problem with activities durations and con- straints. Compared to other problems in the literature, in this case, the ac- tivities have different durations that must be taken into account for proper planning. The planning obtained is suitable for the user’s requirements both in terms of speed in providing results and in terms of simplicity and comprehensibility. In recent years, measures of uncertainty or entropy within evidence theory have again become a topic of interest in the literature. However, this has led to an increase in the already numerous measures of total uncertainty, that is, one that considers both conflict and nonspecificity measures. The research aims to find a unique measure, but none of those proposed so far can meet the required properties. The measures are often complex, and especially in the field of application, it is difficult to understand which is the best one to choose and to understand the numerical results obtained. Therefore, a numerical approach that compares a wide range of measures in pairs is proposed alongside comparisons based on mathematical proper- ties. Rank correlation, hierarchical clustering, and eigenvector centrality are used for comparison. The results obtained are discussed and com- mented on to gain a broader understanding of the behavior of the measures and the similarities and non-similarities between them.