Analiza efektywności metod wyboru zmiennych objaśniających do budowy modelu regresyjnego
Efficiency analysis of chosen methods of explanatory variables selection within the scope of regression model construction

Summary/Abstract: The basic aim of this paper is to compare Akaike’s information criterion and Schwarz’s Bayesian information criterion (BIC), cross entropy and Hellwig’s method within the scope of regression model construction efficiency. The study was based on computer simulations. After generating a dataset with normal distribution, a linear model (true model, which in reality is not known) was built. In the model a response variable is dependent on the chosen variables previously generated. Next, a set of potential explanatory variables was extended and the analyzed methods of model selection were applied. These steps were repeated. Subsequently it was compared how often each of the tested methods indicated the right set of variables, and thus the right model. The methods were compared also on the basis of empirical data.

• Details
• Contents