MULTIPLE REGRESSION.

Abstract

Designing experiments and analyzing data would be simple if every response variable related as a straight line with a single independent variable. Unfortunately, this is not usually the case: In one variable systems, we must be prepared to deal with polynomial, log-normal, Poisson and other nonlinear relationships. More importantly, most of the systems we encounter in chemical engineering have two or more independent variables that affect the response in a nonlinear fashion and also interact with each other. It is not possible to learn much about such a complex system by casually inspecting the data. Instead, one has to rely on formal methods of analysis such as multiple analysis of variance (MANOVA) and multiple regression. This paper discusses the method of multiple regression and shows how this method is used to find the best-fit equation relating a dependent variable with two or more independent ones.