The theory of Loss Functions Is a fundamental part of Statistical Decision Theory and of Pattern Recognition. However It is a subject which few have studied In detail. This thesis is an attempt to develop a simple character recognition process In which losses may be Implemented when and where necessary. After a brief account of the history of Loss Functions and an Introduction to elementary Decision Theory, some examples have been constructed to demonstrate how various decision boundaries approximate to the optimal boundary and what Increase In loss would be associated with these sub-optimal boundaries. The results show that the Euclidean and Hamming distance discriminants can be sufficiently close approximations that the decision process may be legitimately simplified by the use of these linear boundaries. Geometric moments were adopted for the computer simulation of the recognition process because each moment is closely related to the symmetry and structure of a character, unlike many other features. The theory of Moments is discussed, in particular their geometrical properties. A brief description of the programs used in the simulation follows. Two different data sets were investigated, the first being hand-drawn capitals and the second machine-scanned lower case type script. This latter set was in the form of a message, which presented interesting programming problems in itself. The results from the application of different discriminants to these sets under conditions of simple loss are analysed and the recognition efficiencies are found to vary between about 30% and. 99% depending on the number of moments being used and the type of discriminant. Next certain theoretical problems are studied. The relations between the rejection rate, the error rate and the rejection threshold are discussed both theoretically and practically. Also an attempt is made to predict theoretically the variation of efficiency with the number of moments used in the discrimination. This hypothesis is then tested on the data already calculated and shown to be true within reasonable limits. A discussion of moment ordering by defining their re-solving powers is undertaken and it seems likely that the moments normally used unordered are among the most satisfactory. Finally, some time is devoted towards methods of improving recognition efficiency. Information content is discussed along with the possibilities inherent in the use of digraph and trigraph probabilities. A breakdown of the errors in the recognition system adopted here is presented along with suggestions to improve the technique. The execution time of the different decision mechanisms is then inspected and a refined 2-Stage method is produced. Lastly the various methods by which a decision mechanism might be improved are united under a common loss matrix, formed by a product of matrices each of which represents a particular facet of the recognition problem.

Most in-hospital adverse events are attributable to surgical care and of these, clinical decision-making (CDM) errors account for approximately half. The assessment of CDM is integral to establishing competence among surgical trainees. One proposed assessment tool is the script concordance test (SCT), which is based on the dual process theory of CDM, but evidence demonstrating valid results is needed. This thesis collects content and response process validity evidence for the assessment of CDM using the SCT. To gather content evidence, a Delphi technique was conducted with a panel of local general surgeons (n=15) consisting of the top decision-makers voted by staff and resident surgeons. Items achieving consensus were mapped on a table of specifications to determine the breadth of topics covered as defined by the Royal College medical expert competencies in general surgery. The final SCT was administered to 29 residents and 14 staff surgeons and results were analyzed. To gather response process evidence, cognitive interviews were then conducted with ten residents and five staff surgeons based on results of the final SCT. Data from the cognitive interviews were analyzed using a priori deductive codes based on Tourangeau’s cognitive model of response process. The first round of the Delphi yielded agreement ranging from 40-100% and consensus for 21 cases. The 21 cases made up the final SCT and encompassed 13 of the 19 competencies in general surgery. The final SCT reflected a test of the intraoperative management of open, adult general surgery. Notable absent competencies were described by experts to be outside the scope of general surgery, too difficult for the resident level of training, or presenting an unrealistic intraoperative finding. Cognitive interviews demonstrated variability in CDM among test-takers. Consistent with the dual process theory, test-takers relied on scripts formed through past experiences, when available, to make decisions. However, test-takers’ response process was also influenced by issues with respect to their comprehension, recall, and response matching cognitive steps. Due to issues with response matching in particular, when answering an SCT question test-takers indicating different numerical ratings may have the same rationale. The Delphi technique, table of specifications, and cognitive interviews provide validity evidence supporting the SCT for assessing CDM of general surgery trainees. Substantial issues with respect to the numerical rating scale suggests further revisions to the test format are required before consideration of its use in summative assessment.

Cette dissertation est composée de trois contributions à l’intégration de modèles de choix LOGIT dans des problèmes de sélection en variables binaires, ainsi que d’un détail de l’état de l’art à l’intersection de ces deux domaines. Traditionnellement, les problèmes de décision combinatoire considèrent la demande finale comme exogène; d’un autre côté, l’étude des choix des consommateurs est un domaine qui s’est considérablement développé dans la seconde moitié du 20ème siècle, particulièrement avec ce que l’on nomme les Modèles de Choix Discrets. En conséquence, la recherche a vu se multiplier récemment l’intégration de ces modèles de choix discrets pour modéliser la demande finale dans les problèmes de décision. Le décisionnaire peut ainsi prendre en compte les préférences des consommateurs ou utilisateurs finaux ainsi que la façon dont la décision elle-même (par exemple, la conception d’un réseau) peut influencer leurs choix. Comme il est dit dans [13], il s’agit de modéliser l’interaction de l’offre et de la demande, où le domaine de la Théorie des Choix nous fournit des précisions sur la demande, tandis que le domaine de la Recherche Opérationnelle étudie l’optimisation de l’offre. Il existe quantité de modèles de choix, le plus commun et le plus simple étant la régression logistique multinomiale (MNL dans ce texte). Au chapitre 2, nous présentons brièvement les grands modèles de choix les plus utilisés ainsi que leurs hypothèses sous-jacentes. Faisons déjà la remarque qu’un problème d’optimisation intégrant le modèle de choix multinomial est de forme fractionnaire, donc a priori non-linéaire et non-convexe - deux défis dans le domaine de l’optimisation. Un autre défi que pose cette intégration est que les paramètres des modèles de choix ne sont généralement pas aisés à estimer. Dans la section qui suit, nous détaillons les contributions de chaque chapitre. Toutefois, les trois chapitres principaux ont tous comme contribution commune de modéliser un problème nouveau dans le domaine (ou à l’intersection) que nous venons de définir, et de le résoudre par une approche exacte ou une heuristique. / In this dissertation, we study the integration of logit choice models into 0-1 selection models, where a decision maker selects a subset of options (such as facility locations, elements of a network or product features) among a large choice of options. The decision maker offers this subset to her end customers, who make their own choice according to a logit model. The integration of binary variables inside fractional models leads to a computational challenge, which we tackle. The first chapter after the introduction is a literature review where we present the state of research regarding selection models including choice models. In the second chapter, we study a network selection model for healthcare payers in the context of reference pricing, including a robust optimization extension to allow for uncertainty of the choice model parameters. In the third chapter, we study the performance of the first-choice assignment model as a heuristic to solve the hard multinomial logit model. Finally, in the last chapter before the conclusion, we use a nested logit choice model instead of the classic multinomial logit for the location problem. In the nested model with two levels, options are grouped into nests and the choice is made in two steps: first the nest, then the option. As a whole, the dissertation includes contributions on the side of problem modeling, efficient exact solving or metaheuristics, theoretical insights and practitioner insights.

A study has been conducted into improving construction design through the application of the lean thinking paradigm. Its objective was to identify the issues relating to design efficiency and how a lean thinking approach might address these issues. The investigation consisted of examining work already undertaken in the field by other researchers"to identify the state of the art. The change order request system was examined to gain first insights into waste in construction design, and to gauge the size of the opportunity for the application of lean thinking. An Electronic Data Gathering Tool (EDGT) was then developed to allow further exploration of the design decision making process at the system / sub-system level. The EDGT was used on three live construction projects. From the data recorded a design planning tool, Design Decision Planner (DDP), was created to help improve control of the design process and lead to a more standardised approach to construction design. Standardising the approach to product development is an important component of lean thinking. The main recommendations for making construction design lean are: Use DDP to plan and improve control of the design decision making process, assign design responsibility and to make the process more transparent. 2. Measuring progress against planned design is a useful process metric. 3. Improve the designer's cost and programme visibility when choosing between design options. 4. Redefine the role of the quantity surveyor from cost controller to value for money assessor. The role needs to be better integrated into the design process to reach its full potential. 5. Need to develop more rigorous methods of assessing the buildability of design options. This problem could be eased in the short-term by incorporating construction professionals into the early design phases. 6. Designers need to use more process reason drivers when choosing between design options, not just functional criteria. 7. The change order request system could be redesigned to identify the root causes of contract issue design changes and, hence, improve the design decision making process.

Made available in DSpace on 2014-10-09T12:53:08Z (GMT). No. of bitstreams: 0 / Made available in DSpace on 2014-10-09T13:58:31Z (GMT). No. of bitstreams: 1 11656.pdf: 26784123 bytes, checksum: c03f74808c190d27a51bf5f6c9f9702e (MD5) / Tese (Doutoramento) / IPEN/T / Faculdade de Economia, Administracao e Contabilidade, Universidade de Sao Paulo - FEA/USP

Thesis (MTech (Business Administration))--Cape Peninsula University of Technology, 2008 / The purpose of this study is to evaluate the organisational communication and infonnation flow at the SA Cultural History Museum (SACHM) and its decentralised sites with particular reference to its role in decision making and, where necessary, to recommend remedial measures. An extensive literature review was undertaken on organisational communication, infornation and decision making in organisations, with particular reference to the SACHM. The literature review facilitated the development of the research instruments, to assess best practices in terms of the identified key drivers. The research questions are: • The existence of communication practices and processes at SACHM, and • The effective dissemination of information necessary for effective decision making. The statistical results of the empirical study reveal conclusively that the organisational communication, information dissemination and decision making at the SACHM was indeed ineffective. The results indicate the following: • The lack of adherence to shared values and common purpose between managers and staff regarding equity, lack of participation in decision-making, lack of service delivery to local, regional and national communities, lack of mutual respect, lack of dedication, lack of a people centered approach, and finally a lack of interpersonal communication and negotiation; • The lack of application of basic communication fundamentals namely the bottom-up and top-down approaches; • The lack of access to timeous and accurate information to make effective decisions; and • The non-existence of communication processes and practices. Dysfunctional organisational communication and inadequate information flow to both employees and managers, has been identified as having an impact on effective decision making. Arising from the findings of the research, a communication audit is recommended to facilitate the establishment of an effective communication and information framework for the museum.

Thesis (MTech (Business Administration))--Cape Peninsula University of Technology, 2008 / The purpose of this study is to evaluate the organisational communication and infonnation flow at the SA Cultural History Museum (SACHM) and its decentralised sites with particular reference to its role in decision making and, where necessary, to recommend remedial measures. An extensive literature review was undertaken on organisational communication, infonnation and decision making in organisations, with particular reference to the SACHM. The literature review facilitated the development of the research instruments, to assess best practices in terms ofthe identified key drivers. The research questions are: • The existence ofcommunication practices and processes at SACHM, and • The effective dissemination ofinfonnation necessary for effective decision making. The statistical results of the empirical study reveal conclusively that the organisational communication, infonnation dissemination and decision making at the SACHM was indeed ineffective. The results indicate the following: • The lack ofadherence to shared values and common purpose between managers and staff regarding equity, lack ofparticipation in decision-making, lack ofservice delivery to local, regional and national communities, lack of mutual respect, lack ofdedication, lack ofa people centered approach, and finally a lack ofintezpersonal communication and negotiation; • The lack ofapplication ofbasic communication fundamentals namely the bottom-up and top-down approaches; • The lack ofaccess to timeous and accurate information to make effective decisions; and • The non-existence ofcommunication processes and practices. Dysfunctional organisational communication and inadequate information flow to both employees and managers, has been identified as having an impact on effective decision making. Arising from the findings of the research, a communication audit is recommended to facilitate the establishment ofan effective communication and information framework for the museum.

Made available in DSpace on 2014-10-09T12:53:08Z (GMT). No. of bitstreams: 0 / Made available in DSpace on 2014-10-09T13:58:31Z (GMT). No. of bitstreams: 1 11656.pdf: 26784123 bytes, checksum: c03f74808c190d27a51bf5f6c9f9702e (MD5) / Tese (Doutoramento) / IPEN/T / Faculdade de Economia, Administracao e Contabilidade, Universidade de Sao Paulo - FEA/USP

M.Comm.

Submitted in fulfillment of the requirements for the Master's Degree in Information and Communication Technology, Durban, University of Technology, Durban, South Africa, 2017. / This dissertation reports on the development of an algorithm based on an existing matchmaking method to solve diverse decision problems in a multi-person environment. The capacity to effectively achieve a lucrative and accurate decision making is a critical aspect of resource management. But the accuracy of a decision making process can be highly compromised because of the high subjectivity and multiple conflicting attributes that are present in human judgement. multi-person decision making is an effective approach for achieving a lucrative and accurate decision making process. The multi-person decision process has proven to be tedious mainly because the existing multi-person decision making methods are extensions of single decision making methods. This imposes additional computational resources, especially for a large number of decision makers because they aggregate the preferences of several decision makers into a unified format.This work therefore seeks to improve the multi-person decision making process using a matchmaking approach. In doing so, the Hunt ForTune matchmaking algorithm was investigated and improved for this purpose. Thus, the preferences of decision makers for each attribute are collected as an attribute description vector. The attribute, its description vector, flexibility and priority vector are compactly represented as a 4-tuple profile. The improved Hunt ForTune matchmaking algorithm is applied to different sets of multi-person decision problems and offered as an effective way of enhancing decision accuracy. The improved matchmaking decision algorithm is compared with a novel mathematical technique of Hausdorff distance. Results generally show that multi-person matchmaking algorithm is suitable and efficient for diverse decision making in the presence of multiple decision makers. The practical implication of the proposed multi-person matchmaking algorithm for decision making is that it provides a less complicated way to capture and represent the preferences of multiple decision makers irrespective of decision domain. The originality of the work reported in this dissertation is built on a matchmaking algorithm by introducing effective profile representation using vector analysis approach to capture the preferences of multiple decision makers and similarity metrics to provide an efficient and robust way to accurately perform a multi-person decision process. / M