Abstract
Fuzzy linear regression provides means for tackling regression problems lacking a significant amount of data for determining regression models and with vague relationships between the dependent variable and independent variables. In the paper a two-stage construction of a linear regression model is proposed using an enhancement of a minimal vagueness criterion already discussed in fuzzy regression analysis. Two different experiments are set up and used for evaluating both models. The choice of the fuzzy regression model should be made in the context of a practical application.