Les modèles de mélange pour la régression sont utilisés pour modéliser la relation entre la réponse et les prédicteurs, pour des données issues de différentes sous-populations. Dans cette thèse, on étudie des prédicteurs de grande dimension et une réponse de grande dimension. Tout d’abord, on obtient une inégalité oracle ℓ1 satisfaite par l’estimateur du Lasso. On s’intéresse à cet estimateur pour ses propriétés de régularisation ℓ1. On propose aussi deux procédures pour pallier ce problème de classification en grande dimension. La première procédure utilise l’estimateur du maximum de vraisemblance pour estimer la densité conditionnelle inconnue, en se restreignant aux variables actives sélectionnées par un estimateur de type Lasso. La seconde procédure considère la sélection de variables et la réduction de rang pour diminuer la dimension. Pour chaque procédure, on obtient une inégalité oracle, qui explicite la pénalité nécessaire pour sélectionner un modèle proche de l’oracle. On étend ces procédures au cas des données fonctionnelles, où les prédicteurs et la réponse peuvent être des fonctions. Dans ce but, on utilise une approche par ondelettes. Pour chaque procédure, on fournit des algorithmes, et on applique et évalue nos méthodes sur des simulations et des données réelles. En particulier, on illustre la première méthode par des données de consommation électrique. / Finite mixture regression models are useful for modeling the relationship between a response and predictors, arising from different subpopulations. In this thesis, we focus on high-dimensional predictors and a high-dimensional response. First of all, we provide an ℓ1-oracle inequality satisfied by the Lasso estimator. We focus on this estimator for its ℓ1-regularization properties rather than for the variable selection procedure. We also propose two procedures to deal with this issue. The first procedure leads to estimate the unknown conditional mixture density by a maximum likelihood estimator, restricted to the relevant variables selected by an ℓ1-penalized maximum likelihood estimator. The second procedure considers jointly predictor selection and rank reduction for obtaining lower-dimensional approximations of parameters matrices. For each procedure, we get an oracle inequality, which derives the penalty shape of the criterion, depending on the complexity of the random model collection. We extend these procedures to the functional case, where predictors and responses are functions. For this purpose, we use a wavelet-based approach. For each situation, we provide algorithms, apply and evaluate our methods both on simulations and real datasets. In particular, we illustrate the first procedure on an electricity load consumption dataset.

COBOL is an old programming language that, despite its age, is still used in many different areas, such as financial and insurance administration. The Swedish Companies Registration Office, also known as Bolagsverket is no exception. This agency administers information regarding Swedish companies and associations. Although the COBOL programs work, they have many disadvantages. The system is both expensive, inflexible and there are very few new developers available. Bolagsverket would therefore gladly replace COBOL with a more flexible and affordable alternative. This thesis will study and present some solutions and also provide actual examples of how they can be used. Two different products have been studied; Oracle SALT and Micro Focus Visual COBOL. SALT offers a web service interface, which enables the users to access services related to the COBOL programs remotely via the Internet or other networks. Visual COBOL, on the other hand, compiles the code into low-level code and deploys it onto new platforms. Both products can make the use of the code in a more flexible manner. Visual COBOL has, however, more substantial and long-term functionality that Bolagsverket is very interested in pursuing, even though it costs five times as much as SALT. One other main reason is that the second product can drastically reduce the gap between the COBOL programs and the surrounding systems. It is even possible to save money since some components can be reduced or even completely removed. The final conclusion is that neither of the programs offers any pure replacement with regards to COBOL, but they have made the language much more adapted to twenty-first century computer systems. They are therefore considered as a good option instead of rewriting the entire COBOL code into a new language. / COBOL är ett gammalt programmeringsspråk som trots sin ålder fortfarande används inom många olika områden, såsom administration inom finans- och försäkringsbranschen. Bolagsverket i Sverige är inget undantag. Den här myndigheten hanterar information rörande de företag och föreningar som finns i Sverige. Även om COBOL-programmen fungerar som det ska finns det många nackdelar. Det är dyrt, inte särskilt medgörligt och dessutom finns det få nya utvecklare att tillgå. Bolagsverket skulle därför gärna vilja byta ut programmen mot något mer flexibelt och prisvärt alternativ. Det här examensarbetet kommer presentera några aktuella lösningar. Kodexempel kommer också att skapas för att visa hur detta kan användas i praktiken. Två olika produkter kommer att behandlas i rapporten; Oracle SALT och Micro Focus Visual COBOL. SALT erbjuder ett gränssnitt för webtjänster. Det möjliggör att användarna kan komma åt tjänster knutna till COBOL programmen, fast på distans, via exempelvis Internet eller något annat nätverk. Visual COBOL har i sin tur en helt annan lösning, nämligen kompilering av COBOL till ett lågnivåspråk som går att använda på flera nya plattformar. Både produkterna erbjuder stora möjligheter i att göra COBOL mer flexibelt. Visual COBOL har däremot mer omfattande och långsiktiga funktioner som Bolagsverket kan tänkas gå vidare med, trots att det kostar fem gånger så mycket som SALT. En annan fördel med Visual COBOL är att flera delar av arkitekturen kan komma att användas mindre, eller till och med avvecklas helt, och kan således även sänka kostnaderna. Slutsatserna som kan dras är att ingen av produkterna byter ut COBOL helt, däremot kan de göra det gamla programmeringsspråket mer anpassat för 2000-talets moderna datorsystem. De betraktas därmed som fullt dugliga alternativ istället för en total omskrivning av COBOL-koden till ett annat språk.

Testing is an important step in the software development process in order to increase the reliability of the software. There are a number of different methods available to test software that use different approaches to find errors, all with different requirements and possible results. In this thesis we have performed a series of tests on our own mobile application developed for the Android platform. The thesis starts with a theory section in which most of the important terms for software testing are described. Afterwards our own application and test cases are presented. The results of our tests along with our experiences are reviewed and compared to existing studies and literature in the field of testing. The test cases have helped us find a number of faults in our source code that we had not found before. We have discovered that automated testing for Android is a field where there are a lot of good tools, although these are not often used in practice. We believe the app development process could be improved greatly by regularly putting the software through automated testing systems.

This work treats the representation of moving objects and operations over these objects. Intro­duces the support for spatio-temporal data in Oracle Database 10g and presents two designs of moving objects database structure. Upon these designs a database was implemented using the user-defined data types. Sample application provides a graphical output of stored spatial data and allows us to call an implemented spatio-temporal operations. Finally, an evaluation of achieved results is done and possible extensions of project are discussed.

Creating websites providing dynamic services is an extensive process. Intelligent systems are used to create websites with dynamic services. Current intelligent systems are hard to use and configure by the average user. The generated websites are usually custom built to solve one problem and cannot be fully customizable for users on different environments. This thesis presents a technological solution that enables the average user to create websites with dynamic services by providing a number of parameters. The website generator is a web-based application that generates all the components of the website. The components act as portlets and the generated website will be the portal application. The data engine generator creates the website's underlying database. To enable distributed enterprise/business architecture, the data engine generator records the metadata about the database and the website to be generated. The website generator is a cost effective, dynamic, secure, reliable, and scalable solution that outperforms current website generators and portal applications.

The aim of this thesis is to analyze the global market with Business Intelligence, recognize the most significant companies running a business in this area and try to estimate the future trends and development of the market and then to analyze BI tools of two important companies on the market, Microsoft and Oracle. Analysis and comparison of these tools will be done according to specified metrics, which will focus on both technological and business features of the tools. The main contribution should be complex comparison of quality of tools from two big competitors on the market.

Embedded systems are present in many fields of application where failure may be critical. Such systems often possess characteristics that hampers the testing activity, as large amount of produced data and temporal requirements which must be specified and evaluated. There are tools that support the development of models for analysis and simulation still in the design stage. After being evaluated, a model may be used as basis to the implementation. In this case, it is important to ensure that the model is consistent with the specification. Otherwise, a divergence will be propagated to the final code. Therefore, the model must be tested prior to the codification. Simulink is a standard development and simulation tool for models of embedded systems. Its wide application in the industry has promoted the creation of free-software alternatives, as XCos. In the literature, there are researches which seek to improve the testing activity for Simulink-like models. The proposed solutions usually focus on test case selection strategies. However, little efforts have been directed to the oracle problem, that is, the difficulty in evaluating if an execution agrees with the specification. The objective of this doctorate proposal is to provide an oracle generation approach for Simulink-like models which addresses the characteristics previously summarized. Specifically, it is proposed a process, methods, procedures and a tool that enable the partially-automated generation of oracles for such models. As a main contribution, it is expected an improvement in the evaluation process of embedded systems in terms of quality, cost and time / Sistemas embarcados estão presentes em diversas áreas de aplicação em que falhas podem ser críticas. Tais sistemas frequentemente possuem características que tornam a fase de teste particularmente desafiadora, como a produção de grande quantidade de dados e requisitos temporais que precisam ser validados de acordo com a especificação. Existem ferramentas que auxiliam no desenvolvimento de modelos para análise e simulação do comportamento de sistemas embarcados ainda na fase de design. Após ser avaliado, o modelo pode ser usado como base para a implementação. Neste caso, deve-se buscar garantir que um modelo esteja de acordo com a especificação. Do contrário, tal divergência será propagada para a implementação. Portanto, e importante que o modelo seja testado antes da fase de implementação. Simulink e uma ferramenta-padrão de desenvolvimento e simulação de modelos de sistemas embarcados. Sua ampla aplicação na indústria incentivou a criação de alternativas de software livres como XCos. Na literatura, existem pesquisas que visam a aprimorar a atividade de teste de modelos Simulink-like. As soluções propostas geralmente focam em estratégias de seleção de casos de teste. Mas pouco esforço tem sido direcionado ao problema do oráculo, isto e, na dificuldade em avaliar se a execução está de acordo com a especificação. O objetivo desta proposta de doutorado é prover uma abordagem de geração de oráculos de teste para modelos simulink-like que contemple as características previamente resumidas. Especificamente, é proposto um processo, métodos, procedimentos e uma ferramenta que viabilizem a geração parcialmente automatizada de oráculos de teste para modelos Simulink-like. Como contribuição principal, é esperada a melhora da qualidade, custo e tempo do processo de validação de sistemas embarcados suportados por modelagem em Simulink e ferramentas similares

Cette thèse s'inscrit dans les domaines de la statistique non-asymptotique et de la théorie statistique de la sélection de modèle. Son objet est la construction de procédures d'estimation de paramètres en régression hétéroscédastique. Ce cadre reçoit un intérêt croissant depuis plusieurs années dans de nombreux champs d'application. Les résultats présentés reposent principalement sur des inégalités de concentration et sont illustrés par des applications à des données simulées.<br /><br />La première partie de cette thèse consiste dans l'étude du problème d'estimation de la moyenne et de la variance d'un vecteur gaussien à coordonnées indépendantes. Nous proposons une méthode de choix de modèle basée sur un critère de vraisemblance pénalisé. Nous validons théoriquement cette approche du point de vue non-asymptotique en prouvant des majorations de type oracle du risque de Kullback de nos estimateurs et des vitesses de convergence uniforme sur les boules de Hölder.<br /><br />Un second problème que nous abordons est l'estimation de la fonction de régression dans un cadre hétéroscédastique à dépendances connues. Nous développons des procédures de sélection de modèle tant sous des hypothèses gaussiennes que sous des conditions de moment. Des inégalités oracles non-asymptotiques sont données pour nos estimateurs ainsi que des propriétés d'adaptativité. Nous appliquons en particulier ces résultats à l'estimation d'une composante dans un modèle de régression additif.

<p>I denna uppsats undersöks hur Kronofogden har byggt upp sin beställarkompetens inför och under upphandlingen av sitt första affärssystem, samt vilken inverkan behovet av beställarkompetens har haft på upphandlingens upplägg.</p><p>Genom att studera den upphandlingsprocess Kronofogden bedrev vid upphandling av nytt affärssystem och med hjälp av en modell för beställarkompetens identifieras vilka faktorer som påverkat upphandlingen samt vilken inverkan dessa faktorer haft.</p><p>Kronofogden har byggt upp sin beställarkompetens genom att utgå ifrån verksamhetens behov och lärt sig hur ett affärssystem kan tillfredsställa behovet. Därigenom har man förbättrat sina möjligheter att genomföra en framgångsrik upphandling trots att Kronofogden inte har några tidigare erfarenheter av affärssystem. Utifrån den förståelse Kronofogden har byggt upp rörande affärssystem har man valt att begränsa sin lösning för att sedan fortsätta successivt och genom att använda ett testsystem har de fått in en extra säkerhet i upphandlingen.</p>

I denna uppsats undersöks hur Kronofogden har byggt upp sin beställarkompetens inför och under upphandlingen av sitt första affärssystem, samt vilken inverkan behovet av beställarkompetens har haft på upphandlingens upplägg. Genom att studera den upphandlingsprocess Kronofogden bedrev vid upphandling av nytt affärssystem och med hjälp av en modell för beställarkompetens identifieras vilka faktorer som påverkat upphandlingen samt vilken inverkan dessa faktorer haft. Kronofogden har byggt upp sin beställarkompetens genom att utgå ifrån verksamhetens behov och lärt sig hur ett affärssystem kan tillfredsställa behovet. Därigenom har man förbättrat sina möjligheter att genomföra en framgångsrik upphandling trots att Kronofogden inte har några tidigare erfarenheter av affärssystem. Utifrån den förståelse Kronofogden har byggt upp rörande affärssystem har man valt att begränsa sin lösning för att sedan fortsätta successivt och genom att använda ett testsystem har de fått in en extra säkerhet i upphandlingen.