TABLE OF CONTENTS In many quality improvement projects, the linear regression method and the design of experiments method (DOE) are very important tools to model relationships between dependent variables and independent variables. In industrial applications, the dependent variables are often key products or process performance characteristics, and the independent variables are often design parameters or process variables. We often use a P diagram to illustrate this relationship; see Fig. 9.1. For example, in a chemical process, the dependent variables are often the key performance characteristics ( Y 1, Y 2, , Y m) such as yield, purity, etc. The independent variables ( X 1, X 2, , X n) are often process factors such as temperature, pressure, air flow rate, etc. Clearly, if we understand the relationship between these dependent variables and independent variables, we can certainly adjust and control the independent variables, to achieve superior performance. The independent variables are often design parameters or process variables, which can be modeled as dependent variables.
In the design of experiments (DOE) method, we select a set of independent variables and systematically change their settings and measure the values of the dependent variables at these settings. We then construct empirical models relating independent variables and dependent variables. The design of experiments method is a very powerful method. It often effectively finds the causes and relationships among independent variables and dependent variables. It is a powerhouse for quality...
