The concept of classification is used and examined by the scientific community for hundreds of years. In this historical process, different methods and algorithms have been developed and used. Today, although the classification algorithms in literature use different methods, they are acting on a similar basis. This basis is setting the desired data into classes by using defined properties, with a different discourse / an effort to establish a relationship between known features with unknown result. This study was intended to bring a different perspective to this common basis. In this study, not only the basic features of data are used, the class of the data is also included as a parameter. The aim of this method is also using the information in the algorithm that come from a known value. In other words, the class, in which the data is included, is evaluated as an input and the data set is transferred to a higher dimensional space which is a new working environment. In this new environment it is not a classification problem anymore, but a clustering problem. Although this logic is similar with Kernel Methods, the methodologies are different from the way that how they transform the working space. In the projected new space, the clusters based on calculations performed with the Mahalanobis Distance are evaluated in original space with two different heuristics which are center-based and KNN-based algorithm. In both heuristics, increase in classification success rates achieved by this methodology. For center based algorithm, which is more sensitive to new input parameter, up to 8% of enhancement is observed.

No description available.

No description available.

Universidade Federal do Cearà / Esta dissertaÃÃo analisa a existÃncia de risco moral no sistema de saÃde brasileiro a partir de uma anÃlise de estimaÃÃo de dois modelos de matching baseado no propensityscore, um a partir de estratificaÃÃo e outro a partir de uma funÃÃo kernel. A variÃvel de utilizaÃÃo do serviÃo mÃdicoselecionada foi a de consultas mÃdicas e a anÃlise compreendeu os nove estados do Nordeste doBrasil. Osresultados indicam presenÃa de risco moral tanto pela estimaÃÃo a partir de estratoscomo a partir de uma funÃÃo kernel. Para o modelo de estratificaÃÃo o valor estimado da diferenÃa de procura por consultas mÃdicas ficou entre 1,014 e 2,262 vezes, com o estado do RioGrande do Norte apresentando a maior diferenÃa. Jà para o modelo a partir de uma funÃÃo kernel,o valor estimado da diferenÃa de procura por consultas mÃdicas ficou entre 0,866 e 2,242 vezes,com o estado do MaranhÃo apresentando a menor diferenÃa. Ouseja, indivÃduos que possuemplanos de saÃde geralmente buscam maior atendimento mÃdico do que caso nÃo estivessem cobertos por nenhum plano. / This dissertation analised the existence of moral hazard in the brazilian health system using anestimation of two propensity score matching models, one by stratification and the other one usinga kernel function. The variable of healthcare utilization selected was the medical visits and theanalysis was made considering the nine states of Northeast of Brazil. The results indicatedexistence of moral hazard for both stratification and kernel function estimation. For thestratification model the stimated value of the diference in medical visits stayed between 1,014 e 2,262 times, the state of Rio Grande do Norte had the greater diference. For the kernel functionmodel the stimated value of the diference in medical visits stayed between 0,866 e 2,242 times,the state of MaranhÃo had the smallest diference. So, individuals with health insurance use morethe healthcare than if they only have access to public services.

nÃo hà / O objetivo deste trabalho à avaliar a ocorrÃncia de risco moral na demanda por consultas mÃdicas e serviÃos de apÃio ao diagnÃstico e tratamento (exames) no sistema de saÃde suplementar brasileiro. Especificamente, visamos à comparaÃÃo quantitativa da utilizaÃÃo desses serviÃos de saÃde entre dois tipos de planos: com e sem coparticipaÃÃo. Para isso, utilizamos dois mÃtodos de matching baseado no propensity score, onde os estimadores sÃo calculados a partir de grupos ou estratos (Stratification Matching) e a partir de uma funÃÃo densidade (Kernel Matching). No modelo de estratos o efeito estimado da coparticipaÃÃo em consultas mÃdicas ficou entre -0,358 e -0,979 per capita por ano. No modelo a partir de uma funÃÃo Kernel, o efeito da coparticipaÃÃo foi de -0,286 a -1,031 per capita por ano. No caso dos exames, o efeito da coparticipaÃÃo ficou entre -2,965 e -4,652 (Stratification Matching) e -2,621 e -5,057 (Kernel Matching). Em ambos os modelos a regiÃo Nordeste apresentou o maior efeito de coparticipaÃÃo, tanto para consultas quanto para exames, enquanto a regiÃo Sul teve o menor efeito. Os resultados alcanÃados com o trabalho evidenciam a ocorrÃncia de risco moral, ou seja, a utilizaÃÃo de serviÃos de saÃde à sensÃvel à imposiÃÃo de mecanismos de regulaÃÃo, onde indivÃduos com coparticipaÃÃo demandam consultas mÃdicas e exames em quantidade inferior aqueles sem coparticipaÃÃo. / The objective of this study is to evaluate the occurrence of moral hazard in the demand for medical and support services for diagnosis and treatment (tests) in the Brazilian supplementary health system. Specifically, we aim at the quantitative comparison of the use of health services between two types of plans: with and without coparticipaÃÃo. For this, we use two methods of matching based on propensity score, where the estimators are derived from groups or strata (Stratification Matching) and from a density function (Kernel Matching). In the model of strata the estimated effect of coparticipaÃÃo medical appointments was between -0.358 and -0.979 per capita per year. In the model from a kernel function, the effect of coparcenary was -0.286 to -1.031 per capita per year. In the case of examinations, the effect of coparcenary was between -2.965 and -4.652 (Stratification Matching) and -2.621 and -5.057 (Kernel Matching). In both models the Northeast had the highest effect coparcenary, both for consultations and for examinations, while the South had the smallest effect. The results achieved through the study show the occurrence of moral hazard, ie the use of health services is sensitive to the imposition of regulatory mechanisms where individuals coparticipaÃÃo require medical examinations and in quantities not exceeding those without coparcenary

Dans cette recherche, on s’intéresse à l’étude asymptotique des méthodes de points intérieurs pour la programmation linéaire. En se basant sur les travaux de Schrijver et Padberg, nous proposons deux nouveaux pas de déplacement pour accélérer la convergence de l'algorithme de Karmarkar et réduire sa complexité algorithmique. Le premier pas est une amélioration modérée du comportement de l'algorithme, le deuxième représente le meilleur pas de déplacement fixe obtenu jusqu'à présent. Ensuite nous proposons deux approches paramétrées de la l'algorithme de trajectoire centrale basé sur les fonctions noyau. La première fonction généralise la fonction noyau proposé par Y. Q. Bai et al., la deuxième est la première fonction noyau trigonométrique qui donne la meilleure complexité algorithmique, obtenue jusqu'à présent. Ces propositions ont apporté des nouvelles contributions d'ordre algorithmique, théorique et numérique. / In this research, we are interested by asymptotic study of interior point methods for linear programming. By basing itself on the works of Schrijver and Padberg, we propose two new displacement steps to accelerate the convergence of Karmarkar's algorithm and reduce its algorithmic complexity. The first step is a moderate improvement of the behaviour of this algorithm; the second represents the best fixed displacement step obtained actually. We propose two parameterized approaches of the central trajectory algorithm via a kernel function. The first function generalizes the kernel function given by Y. Q. Bai et al., the second is the first trigonometric kernel function that gives the best algorithmic complexity, obtained until now. These proposals have made new contributions of algorithmic, theoretical and numerical order.

No description available.

The goal of this master's thesis is to acquaint with a problem of a knowledge discovery and objectrelational data classification. It summarizes problems which are connected with mining spatiotemporal data. There is described data mining kernel algorithm SVM. The second part solves classification method implementation. This method solves data mining in a Caretaker trajectory database. This thesis contains application's implementation for spatio-temporal data preprocessing, their organization in database and presentation too.