Estudo do modelo de famÃlias de distribuiÃÃes de probabilidade baseado em programaÃÃo matemÃtica

AUTOR(ES)
DATA DE PUBLICAÇÃO

2007

RESUMO

Since immemorial times, man has learned to deal with uncertainty in trying to prevent losses derived from unexpected factors. Several probability theories were proposed in such a search, KolmogorovÅs having been the most used. However, it fails to serve a number of situations. A great deal has been done in order to work out the situations where the classic probability fails, such as the Choquet capacity, the theory of evidence by Dampster-Shafer, upper and lower probabilities, among others. This work gives continuity to developing the model of representation and calculation of the uncertainty based on the linear programming introduced by Campello de Souza (1993), whose latest results are in Campello de Souza (2007). The model uses probability distribution families to represent and quantify uncertainty. Some applications of the model in the elicitation of specialistsÅ knowledge have been made and a new measurement for the specialistsÅ inferential skills has been proposed. The model was used to work with the statistical inference when data were scarce, working on estimates of averages of probability distributions, which were compared to the relative frequency method. Decidability was used to measure the association between two random variables, which was discovered to be linearly correlated to Pearsonâs correlation

ASSUNTO(S)

incerteza probabilities families programaÃÃo matemÃtica mathematical programming eduÃÃo uncertainty famÃlias de probabilidade engenharia de producao elicitation

Documentos Relacionados