Files
Abstract
We consider interval-valued data that are commonly observed with the advanced technology in the current data collection processes. Interval-valued data are collected as intervals while classical data are formatted as single values. In this thesis, we are particularly interested in regression analysis. This thesis starts with a brief review on the existing methods for the regression analysis of interval-valued data. Then, we propose a new approach to fit a linear regression model to interval-valued data using bootstrap. The proposed method enables one to do statistical inferences concerning regression coefficients where most of the existing methods fail to provide. The proposed and existing methods are applied to the real and simulated data and their performances are compared each other.