Previous Page

|

Next Page

Points to Remember

• You can create different folders for organizing your designs in ADX.
• Once responses are entered into a design, the design is frozen and can no longer be changed.
• ADX uses coded values (-1 for low and 1 for high) for factors when fitting.
• The Explore Data window provides a variety of graphical displays for examining the structure of your data prior to model fitting.
• Main-effects plots provide a way to look at the effects of individual factors on the response.
• Interaction plots indicate the presence and strength of two-factor interactions.
• The Zoom feature provides more detail for individual cells within the interaction matrix.
• Factorial plots help you to visualize the factorial structure in tree form.
• The Effect Selection window provides a wealth of methods for automatically selecting significant effects in a fitted master model.
• ADX uses ANOVA as the automatic effect selection method for a full factorial model with degrees of freedom for estimating error. ADX uses Lenth's method for saturated factorial designs.
• The Effect Selection window offers a variety of graphical methods for displaying significant effects, including normal and Pareto plots.
• ADX follows the principle of model hierarchy for building the predictive model. You can override this principle.
• The Prediction Profiler is an interactive tool for exploring the effect of changing factor settings on predicted values.
• You can augment the Prediction Profiler with desirability functions that you can use to find those factor settings that maximize the overall desirability.