Thesis (M.Sc.) - Cairo University - Faculty of Graduate Studies for Statistical Research - Department of Statistics and Econometrics

Model selection methods in regression analysis have statistical value, especially the case of the models with multiple independent variables and then-recent developments in model selection methods to extract useful information from large databases (Big Data) in all fields. However, traditional statistical methods are unable to manage this bases of big data. Extracting useful information from these complex and informative rules has become a major challenge.The summary of this thesis is to compare between classical variable selection methods like ordinary least square (OLS), least absolute shrinkage and selection operator (LASSO), random forests, principle component analysis with neural network and two proposed methods are random forests with neural network and LASSO with neural network in Monte Carlo simulation study and application in real data with criteria (MSE, MAE and RMSE)

Issued also as CD