Thesis (M.Sc.) - Cairo University - Faculty of Economics and Political Science - Department of Statistics

Recently, quantile regression has received great attention in data analysis because it captures the effect of the independent variables on the whole distribution of the dependent variable, unlike traditional linear regression which provides only one estimate as a representative of this effect for the entire distribution. In this thesis, a modified weighted goal programming model is proposed to obtain the estimates of the quantile regression parameters with less computational effort than the original model of the quantile regression. The proposed model saves more than half the CPU time used for computing quantile regression estimates compared to the original model. A simulation study is performed to assess the performance of the proposed model. Also, the quantile regression is applied to the students{u2019} scores dataset of Faculty of Economics and Political Science at Cairo University to determine the effect of the COVID-19 pandemic on the students{u2019} performance

Issued also as CD