Taguchi designs

What is Taguchi method?

The Taguchi method is a method introduced by Genichi Taguchi (Genichi and Wu, 1980) which is a method of design of experiment providing an improvement to full and fractional factorial designs.

Two steps are needed to apply that method:

• A step of generation of the Taguchi design during which it is possible to choose from a list of designs according to the number of factors and the levels of these.

• Once the Taguchi design is obtained, you can analyze it to identify the control factor parameters that minimize the variation of the response. This second step is achievable with the tool Analysis of a Taguchi design.

Taguchi design options within XLSTAT

Depending on the number of factors and their levels, XLSTAT offers a list of appropriate designs. These designs have a particular notation defined such as: L(nbRuns)(nbLevel^nbFactor).

Where nbRuns = number of runs, nbLevel = number of levels for each factor, and nbFactors = number of factors.

If the notation is of this form: L(nbRuns)(nbLevel^nbFactors), the design contains mixed-levels factors.

For example, a L9(3^3) design is a design having 9 runs and 3 factors at 3 levels. A L18(3^3 6^1) design is a design having 18 runs, 3 factors with 3 levels and 1 factor with 6 levels.

Taguich design output within XLSTAT

Variable Information: This table summarizes all information about the selected factors.

Design of experiment: This table shows the Taguchi design. Empty columns are used to fill in the responses.

Start the analysis: Once the responses columns are filled, you can click on the "Run the analysis" button to open the prefilled dialog box allowing to perform the analysis of the Taguchi design.

Coded design: This table displays Taguchi's coded design, containing 1, 2, 3, ... rather than the categories of your factors.