LEAST SQUARES CALCULATIONS WITH GAMS

This document show how different type of regression models can be solved with GAMS. 1. Linear Least Squares 1.1. OLS is an optimization problem. Ordinary Least Squares (OLS) is a technique to estimate parameters in a linear statistical model: (1) y = Xβ + where y is the dependent (endogenous) variable (stored as an (n× 1) vector), and X is an (n× k) matrix of k independent (exogenous) variables. is an error term. We assume that E( ′ ) = σIn, i.e. the different i’s are independent. We can estimate β by the optimization model: