B1790
Title: Aspects of statistical inference on interval-valued data
Authors: Conceicao Amado - Universidade de Lisboa (Portugal) [presenting]
Catarina Rodrigues - Universidade de Lisboa (Portugal)
Abstract: In traditional statistics, it is assumed that the precise values of the associated quantities are known. However, in some situations, owing to the maintenance of confidentiality or the accessibility of data, there are intervals containing these values. Symbolic data is a concept for dealing with this data. Although various methods have been proposed to handle interval data in symbolic data analysis, including regression models, principal component analysis, and clustering, only a limited number of these approaches have considered issues related to inference. Existing techniques are discussed for hypothesis tests on interval data for the mean, and hypothesis tests are developed that take the interval centers into account as well as a random choice of a value in the interval. In addition, using bootstrap hypothesis testing is proposed for this type of symbolic data. In a numerical experiment, the efficacy of various strategies is compared.
