Increment<=0

Factor Dependency used to solve critical values

Contour Plots

Too many terms in effect 

Only two terms may be checked

Go

By

To

From

Check two factors, edit grid/scale values.

Contour Plot Specification

Predicted Value at Solution

Critical values outside data range

Loss of Accuracy in eigenvectors

Couldn't Solve Critical Values. Singular.

Effects more than quadratic not allowed

Term should be continuous

Only linear and quadratic allowed

Response: 

Interaction Profiles

Cube Plot

Cube plots only supported for 2 to 6 terms

Box-Cox Transformations

SSE

Lambda

Cannot do Box-Cox because of negative Y values

Cannot do Box-Cox because of no error term

Desirability Values must be between 0 and 1

Desirability

Specify term value

Prediction Profile

Screening Fit

No Multiple Nominal Y's allowed.

Y Values must be in ascending order

Screening Fit does not distinguish Random Effects.



Posterior Prob that sample is uncontaminated:

Posterior

Prior

Std Err Est

Help

Go

DFE

Std Err Scale

K Contam

Prior Prob

Estimate

Response: 

Transformed Parameter Estimates



The estimates have the same variance.

The estimates have different variances and need scaling.

The estimates are not correlated.

The estimates are correlated and need a transformation.

Not full rank. Cannot do screening.

Must be at least 6 parameters

Specify Prior Contamination Coef. K, and Click Go

Bayes Plot

Scaled Estimate

Pareto Plot of Scaled Estimates

Transformation

Transformation of Estimates

Correlation

Correlation of Estimates

Effect Screening

Red line is RMSE.

Blue line is Lenth's PSE.

Normal Plot

Normal Quantile

Normalized Estimates

Lenth PSE

Normalized

Scaled

Original