Simplex Tab

Select the Simplex tab of the Response Optimization Startup Panel to access the options described here. This tab is available only if the Simplex algorithm option is selected on the Quick tab.

Element Name Element Type Description
Combine models Check box Select this check box to optimize the response of the predictive models on a collective (i.e., ensemble) basis. If it is not selected, the Simplex algorithm optimizes the predictive models on a standalone basis, i.e., treats each model separately. This option is available only when more than one predictive model is loaded. By combining your models, you effect use them as an ensemble
Number of iterations Simplex is an iterative algorithm that requires a sufficient number of iterations to converge. Since this sufficient number of iterations is not known a priori , it is usually recommended that you set the number of iterations to as high as possible. Use this option to set the maximum number of iterations the Simplex is allowed to perform.
Note: In reality, the Simplex algorithm may take a substantially lesser number of iterations to converge. This, however, depends on the stopping criteria
Stopping at To shorten the computational time, whenever possible, you can use this option to specify a minimum change in the response value that must be made in order for the search to be terminated before the maximum number of iterations is reached; if the rate of change drops to this level or below, the search is terminated. The default is 0.0001, but the optimal value may vary from one problem to another
Search settings Button Click this button to display a STATISTICA , which you can use to modify the defaults for the start and step values of the Simplex algorithm, as well as the minimum and maximum range of the independent variables that should be covered by the Simplex.

These setting may well determine the actual number of iterations (see Number of iterations, above) that the Simplex requires before a solution is reached. The closer the Simplex starting point to the solution, the faster it converges. The Search settings list is located just below the Search settings button. It provides you with a quick way of viewing the current settings of the Simplex algorithm

Note: To exclude a continuous independent variable from the search, you can set its step size to zero, in which case the given starting value is used in all stages of the algorithm
Search settings Button Click this button to print the current settings of the Simplex algorithm to a standard STATISTICA Spreadsheet