Home / Design of Experiments (DOE)

Design of Experiments (DOE)

The DOE tools build an experiment, fit a model to the results, and then let you read that model. The three groups on the QXL DOE New ribbon tab follow that order: Design, Analysis and Options.

• The DOE Ribbon, every command on the tab and what it opens.

Design

Build a design sheet with the factors, the runs and the empty response columns.

The Design Wizard asks what you have and picks a design for you. Choosing a design yourself goes through Create Design, whose entries are grouped as Modeling, Screening, Special and Import:

• Modeling: 2-Level Factorial, 3-Level Factorial, N-Level Factorial, Central Composite (CCD), Box-Behnken
• Screening: Plackett-Burman, Taguchi
• Special: Setup Historical Analysis and Custom Design, which are two entries onto the same dialog, and Optimal Design
• Import: From DOE PRO XL

Modify Design changes a design that already exists: its factors, its outputs, its replicates and blocks, its coding, and the order of its runs.

Analysis

Fit a model, then read it.

• Regression, and the three engines it uses: Ordinary Least Squares, Binary Logistic and Nominal Logistic
• Uncoded Coefficients, the same fit reported in the factors' own units
• Optimize, search the fitted model for settings that meet a goal
• Charts, ten ways to look at the fit or the raw data

Options

• DOE Options, the settings that change what a regression reports and how designs are built

Additional Information

• Screening vs Modeling
• Blocking
• Folding
• Power and Sample Size
• Categorical Inputs
• Types of Outputs
• The S-Hat Model

The mathematics

• DOE Math: design coding, the prediction equation, the sums of squares behind the ANOVA table, and the power and axial value calculations