Factor selection in screening designs

Kristína Sakmárová, Charles University

Co-authors: Arnošt Komárek, Charles University; Martin Otava, Janssen-Cilag s.r.o

Abstract: Throughout the manufacturing process in pharmaceutical research, series of factors can have impact on the quality of the product and there can be numerous interactions between these factors. Our aim is to choose the factors that possibly affect the product the most. To be able to choose as few factors which have the effect on the product as possible, penalization of the regression coefficients needs to be applied. We compare several methods of variable selection in context of data from designed experiments and show that the robustness of the methods does not necessarily need to lead to robustness in selection.