The
Cornerstone™ Experimental Design
module brings DOE to the mainstream, by
making DOE capabilities available to
users of all levels through an
easy-to-use, graphical interface to the
most commonly used DOE designs. The
Experimental Design module provides a
four-step guide to performing successful
designed experiments. In addition to
guidance and ease of use, the new
Cornerstone module provides all the
functionality you need to structure
designed experiments.
Data
from a designed experiment provides a
solid foundation for analysis, because a
designed experiment requires you to
select factor settings using statistical
principles and collect data that is
balanced and has known statistical
properties. The Experimental Design
module suggests several different
designs that you can use to collect
balanced data while reducing the number
of data points you need to collect.
The
Experimental Design module provides a
graphical step-by-step guide to creating
designed experiments:
1.
Define - you define the factors (inputs)
and responses (outputs) of the process
or product.
2.
Design - you select a model and a
design.
3.
Enter data - you carry out the design,
generate a worksheet, and enter response
values.
4.
Analyze - you perform analysis of the
experimental data.
Defining
Variables
The
Define menu makes specifying factors and
responses an easy task. A Quick
Definition menu option displays a dialog
box for quickly defining factors and
responses. You can display a summary of
the factors and the responses you have
defined. In addition, you can display a
dialog box for defining the constraints
that the factors must satisfy.
The
system measures a multiple measurement
response multiple times at the same
factor settings. You enter values for
each measurement, and the system
calculates summary statistics such as
the mean and standard deviation of the
multiple response values at each factor
setting. During analysis, you have a
choice of using the raw data as
responses in your regression, of using
the calculated summary statistics, or
both.
Choosing
a Model and Design
In step
two, the system guides you in selecting
the model you want to fit and the design
you will use to fit that model. You
begin by entering the model type as
Linear, Interactions, Quadratic or
User-Defined. Using the Interaction
button, you have the option of selecting
pairs of factors to appear in two-factor
interaction terms. The Quadratic button
allows you to choose which continuous
factors will appear in squared terms.
The system keeps track of the number of
terms that are currently defined with
each model type.
Entering
Data
There
are 2 ways to enter response data;
a spreadsheet or a checklist. The
spreadsheet shows factor settings based
on the design you have selected. Empty
columns are set up for response values.
You can add runs to a spreaedsheet by
adding rows as long as you specify a
value for each factor.
Analysing
Data
In this
step the system creates a Cornerstone
dataset and invokes the Cornerstone
Regression module to fit the design
model you have selected. There are
several new options for regression
analysis - effects, interactions,
co-efficients table, residuals and
prediction.
The
Experimental Design module allows you to
design of experiments to obtain maximum
information from minimum analysis steps
