Μέθοδοι επιλογής βέλτιστου συνόλου ανεξάρτητων μεταβλητών σε μοντέλα γραμμικής παλινδρόμησης
Τροχοπούλου, Βασιλική Μ.
In linear regression, one of the most important problems is the selection of a subset of independent variables that are available, so that on one hand the prediction of the dependent variable is cost effective, and on the other hand the efficiency loss experienced in the prediction model is as small as possible. This problem has many practical applications in various fields, such as social sciences, economics, meteorological phenomena and many others. In the literature a variety of methods and optimality criteria have been suggested for selecting the appropriate set of variables. In this dissertation we present the most popular techniques which lead to the selection of the optimal set of independent variables, in order to predict a dependent variable through a linear model.