Programming as methodology for parameter estimation in multiple linear regression model

This work results from the concrete need for the mathematical- statistical basis of a business decision on investment in a small enterprise. The interdependence of variables is estimated by the standard statistical way and subsequently parameters are estimated by linear programming by the l infinity. norm. However, besides the precise estimation of the relevant interdependence in the linear multiple regression model the manager also requires an estimation of parameter values resulting to some extent from the underestimation of the total income due to turbulent conditions of transitional economy. This additional requirement was met by evaluating the model using goal programming by the l(1) norm, with introduction of the corresponding weights.