Interval Arithmetic for the Evaluation of Imprecise Data Effects in Least Squares Linear Regression

In this paper, we present a new approach for evaluating the effects of data imprecision on regression analysis. In particular we define the linear regression model in terms of interval valued variables and we describe the conditions under which this model admits interval-valued solutions. We refer to the Interval Arithmetic in order to treat the interval-valued variables as a whole. An application points out the methods capabilities.