Linear & Quadratic Models


In this kind of experiment, you measure how the response (the dependant variable) changes when the factor (the independent variable) changes from a low to a high level. You also measure the interaction between factor. The regression formula for this kind of model is in the image1.

Image1 - Linear regression example
Image1 – Linear regression example


Balanced & Orthogonal Designs

A balanced design is one where each factor (independent variable) have the same number of observation of the otherr factors.

An Orthogonal design is one where each factor can be evaluated independently from other factors. In other words, you can determine each factor independently because you have no interaction with the other.


Fit, Diagnose Model, and Center Points

The center points are the center level of each factors. For example if we have height 60,70,80 the center point is 70.

The center point are used to:

  • To measure the process stability and variability;
  • To check for curvature: in this case, maybe we look that from the minimum and the maximum value there is a curve and not only a linear correlation;

This center point in the Design of the experiment looks like an extra run of the test, so we need to consider the additional cost if we want to use them.

Share on: