In many areas of Artificial Intelligence (AI), we are interested in helping people make better decisions. This help can result in two advantages. First, computers can process large amounts of data and perform quick calculations, leading to better decisions. Second, if a user does not have to think about some decisions, they have more time to focus on other things they find important. Since users' preferences are private, in order to make intelligent decisions, we need to elicit an accurate model of the users' preferences for different outcomes. We are specifically interested in outcomes involving a degree of risk or uncertainty. A common goal in AI preference elicitation is minimizing regret, or loss of utility. We are often interested in minimax regret, or minimizing the worst-case regret. This thesis examines three important aspects of preference elicitation and minimax regret. First, the standard elicitation process in AI assumes users' preferences follow the axioms of Expected Utility Theory (EUT). However, there is strong evidence from psychology that people may systematically deviate from EUT. Cumulative prospect theory (CPT) is an alternative model to expected utility theory which has been shown empirically to better explain humans' decision-making in risky settings. We show that the standard elicitation process can be incompatible with CPT. We develop a new elicitation process that is compatible with both CPT and minimax regret. Second, since minimax regret focuses on the worst-case regret, minimax regret is often an overly cautious estimate of the actual regret. As a result, using minimax regret can often create an unnecessarily long elicitation process. We create a new measure of regret that can be a more accurate estimate of the actual regret. Our measurement of regret is especially well suited for eliciting preferences from multiple users. Finally, we examine issues of multiattribute preferences. Multiattribute preferences provide a natural way for people to reason about preferences. Unfortunately, in the worst-case, the complexity of a user's preferences grows exponentially with respect to the number of attributes. Several models have been proposed to help create compact representations of multiattribute preferences. We compare both the worst-case and average-case relative compactness.

In many areas of Artificial Intelligence (AI), we are interested in helping people make better decisions. This help can result in two advantages. First, computers can process large amounts of data and perform quick calculations, leading to better decisions. Second, if a user does not have to think about some decisions, they have more time to focus on other things they find important. Since users' preferences are private, in order to make intelligent decisions, we need to elicit an accurate model of the users' preferences for different outcomes. We are specifically interested in outcomes involving a degree of risk or uncertainty. A common goal in AI preference elicitation is minimizing regret, or loss of utility. We are often interested in minimax regret, or minimizing the worst-case regret. This thesis examines three important aspects of preference elicitation and minimax regret. First, the standard elicitation process in AI assumes users' preferences follow the axioms of Expected Utility Theory (EUT). However, there is strong evidence from psychology that people may systematically deviate from EUT. Cumulative prospect theory (CPT) is an alternative model to expected utility theory which has been shown empirically to better explain humans' decision-making in risky settings. We show that the standard elicitation process can be incompatible with CPT. We develop a new elicitation process that is compatible with both CPT and minimax regret. Second, since minimax regret focuses on the worst-case regret, minimax regret is often an overly cautious estimate of the actual regret. As a result, using minimax regret can often create an unnecessarily long elicitation process. We create a new measure of regret that can be a more accurate estimate of the actual regret. Our measurement of regret is especially well suited for eliciting preferences from multiple users. Finally, we examine issues of multiattribute preferences. Multiattribute preferences provide a natural way for people to reason about preferences. Unfortunately, in the worst-case, the complexity of a user's preferences grows exponentially with respect to the number of attributes. Several models have been proposed to help create compact representations of multiattribute preferences. We compare both the worst-case and average-case relative compactness.

Sammanfattning Beteendeekonomi var fram till år 1979 ett forskningsämne som saknade större motsättningar. Sedan 1700-talet var den allmänna uppfattningen att de beslut som individer fattade under risk var baserade på ett rationellt beteende. Daniel Kahneman och Amos Tverskys åsikt var polär mot den tidigare forskningen och baserat på deras kritik mot föregående studier inom beteendeekonomi presenterade de år 1979 prospektteorin, en teori som senare renderade i nobelpriset. Därefter har teorin utvecklats och år 1992 publicerade Tversky och Kahneman den kumulativa prospektteorin. Den kumulativa prospektteorin (1992) baseras på att individer frångår objektiva sannolikheter och istället utgår beslut från subjektiva preferenser och därav ett irrationellt beteende. Kahneman och Tversky ansåg att rationella individer inte alltid fattar beslut baserat på vilket alternativ som genererar den högsta nyttan utan tidigare erfarenheter och upplevelser resulterar i att individer agerar annorlunda. Ett flertal studier har funnit empiriskt bevis för att den kumulativa prospektteorin är applicerbar på investerare, däribland på förvaltare inom fonder samt inom private banking. Denna studies syfte är att med hjälp av tolv kvalitativa intervjuer erhålla en djupare förståelse huruvida den kumulativa prospektteorin är applicerbar på svenska fondförvaltare med inriktning på företagsobligationer. Samtidigt som allmänheten enligt de intervjuade förvaltarna tenderar att ha bristfälliga kunskaper gällande risker associerade till företagsobligationer anser många journalister, bland annat på grund av de förväntade räntehöjningarna, att obligationsmarknaden befinner sig i en bubbla. Detta gör företagsobligationsmarknaden intressant att undersöka. Studiens slutsats är att förvaltarna, i likhet med den kumulativa prospektteorin, agerar irrationellt vid investeringsbeslut. Detta på grund av att förvaltarna ger indikationer på att de inte enbart investerar i de företagsobligationer som genererar den högsta nyttan, det vill säga avkastning, utan tar stor hänsyn till risker kopplade till företagsobligationer. I likhet med teorin tenderar förvaltarna att hantera likviditetsproblematiken och kreditrisken i enlighet med den kumulativa prospektteorin. Vidare är studiens slutsats att förvaltarna, i kontrast till den kumulativa prospektteorin, övervärderar en redan hög sannolikhet för att ränte- och inflationsrisken ska påverka fonderna negativt. Dessutom ges indikationer att förvaltarna, i likhet med teorin, agerar riskavert mot vinster, men i kontrast till teorin, agerar de också riskavert mot förluster. Detta stöds bland annat genom att majoriteten av förvaltarna agerar med en hög grad av försiktighet samt deras bemötande av kreditrisk.

This thesis analyzes factors affecting adoption of integrated pest management (IPM) techniques by Kenyan vegetable farmers, including the role of their risk preferences. It also analyzes factors affecting their pesticide applications and expenditures. A survey was administered to 450 Kenyan vegetable growers to identify their pest management practices, and a behavioral experiment was run to elicit their risk preferences utilizing. Cumulative Prospect Theory. Loss aversion was found to be correlated with higher likelihood of IPM adoption while risk aversion was associated with higher pesticide application rates and expenditures. The influence of IPM adoption on pesticide use differed by IPM technique. / Master of Science / Integrated Pest Management (IPM) techniques can improve small holder farmers' livelihoods by lowering production costs and decreasing dependence on chemical pesticides. Even though some IPM techniques have been available to Kenyan vegetable farmers since the 1990's, IPM adoption remains relatively low while chemical pesticide use remains high. A farm-household survey and behavioral experiment were conducted to identify factors that influence farmer decisions to adopt IPM and to apply pesticides. Factors that influence IPM adoption were found to differ from those that influence pesticide decisions. Furthermore, IPM adoption by Kenyan farmers does not decrease use of chemical pesticides for all IPM techniques.

Cette thèse a pour objectif d’évaluer et de comparer la performance des stratégies d’assurance de portefeuille pour tenter de définir quelles stratégies doivent être privilégiées par les investisseurs. Nous comparons de nombreuses stratégies d’assurance (OBPI, CPPI, put synthétique et Stop-loss) entre elles mais également avec quelques autres stratégies de référence. Nous utilisons différents critères de comparaison qui comprennent: 1. Les distributions de pay-off, le niveau de protection, la dominance stochastique et le coût d’assurance dans différentes conditions de marché identifiées par des modèles à changements de régime markovien. 2. Les mesures de la performance ajustée au risque qui peuvent refléter les préférences des investisseurs vis-à-vis du risque et de la rentabilité. 3. Les préférences des investisseurs en intégrant la théorie cumulative des perspectives (TCP). Nos résultats semblent mettre en évidence une dominance des stratégies CPPI dans la majorité des cas et pour la majorité des critères de comparaison. / This thesis is set out with the objective of evaluating and comparing the performance of portfolio insurance strategies. We try to figure out when and why one portfolio insurance strategy should be preferred by investors in practice. To meet this objective, main portfolio insurance strategies (OBPI, CPPI, Synthetic put and Stop-loss) are compared relatively to each other and to some benchmark strategies. Portfolio insurance strategies are applied within different implementation scenarios and compared according to various criteria that include:1. The payoff functions, stochastic dominance, the level of protection and the cost of insurance under bull and bear market conditions. 2. Various risk adjusted performance measures that reflect different investors’ preferences toward risk and return. 3. The preferences of investors who act according to cumulative prospect theory (CPT). Our results reveal a dominant role of CPPI strategy at the majority of cases and according to the majority of comparison criteria.