Optimal Design Result: Mixture (No Intercept) - Quick Tab
Select the Quick tab of the Optimal Design Result: Mixture (No Intercept) dialog box to access the options described here.
- Display design
- Click the Display design button to display a spreadsheet with the selected points. The way in which these points are displayed is affected by the settings of the options on the
Display design tab. To save the design, select the spreadsheet and use the Save or Save As options on the File menu.
The second column of the spreadsheet contains the prediction variances. For design point i, this value is computed as (see Mitchell, 1974b):
si2 = f(xi)'*(X'X)-1*f(xi)
where X stands for the design matrix, and f(xi) is the vector of codes associated with the factor effects in the current model for design point i.
- Order of runs
- The choice of options in the Order of runs group box affects the order of runs displayed in the spreadsheet (via option Display design, see above).
- Standard order
- Select the Standard order option button to display the design in standard order (the order in which it appears in the candidate list).
- Random
- To randomize the order of runs, select the Random option button. For standard experiments, it is common practice to randomize the runs in order to avoid any systematic biases due to fluctuations in the production process across runs.
- Summary: Efficiency measures
- Click the Summary: Efficiency measures button to produce a spreadsheet containing the summary of efficiency measures. For a description of these efficiency measures, see Efficiency Measures for D- and A-Optimal Designs. Note that the only G-efficiency measures are reported for a mixture model (see the D- and A-Optimal Designs - Options tab). For a discussion of the different optimality criteria, refer to the Introductory Overview.
- Triangular scatterplot
- Click the Triangular scatterplot button to display a triangular scatterplot of the design points. If there are more than three factors in the current design, after clicking this button, the Select factors for 3D Scatterplot dialog box will be displayed. In this dialog box, select the three factors for the 3D scatterplot. Note that this scatterplot shows not only points for which the sum of the components is equal to the mixture total, but also "projected" points. If there are only three factors in the design, the sum of the component values are, of course, always equal to the mixture total. However, if there are more than three variables in the design, the scatterplot also shows those points where the values for the components in the plot is not equal to the mixture total (but greater than 0). The values for those points are proportionately adjusted in the plot (maintaining the ratio of the component values), so that they add to the mixture total.
Copyright © 2021. Cloud Software Group, Inc. All Rights Reserved.