Response Optimization Startup Panel

Ribbon Bar: Select the Data Mining tab. In the Tools group, click Optimizations and select Response Optimization option to display the Response Optimization Startup Panel.

Classic menus : On the Data Mining - Process Optimization submenu, select Response Optimization for Data Mining Models option to display the Response Optimization Startup Panel.

The Startup Panel can contain four different tabs:
  • Quick
  • Simplex: For the simplex algorithm
  • Grid: For the grid algorithm.
  • Random: For the random algorithm
  • Model Exploration

For more information, see also

Element Name Description
Load models Click this button to load the selected PMML input model files. More than one model file can be loaded, and in this case, all models are applied to the data specified as the current active data set
Variables Click this button to print a spreadsheet of the model variables including the dependent (response) and independent variables. The information contained in this spreadsheet includes the numerical nature, i.e., continuous or categorical, of each variable and some descriptive statistics such as mean and variance as well as the minimum and the maximum
Optimize Click this button to start optimizing the response of the loaded model(s) using one of the optimization techniques provided in the Algorithm group box of the Quick tab. While the optimization is in progress, a progress bar is displayed containing information relating to the progress of the algorithm, such as the name of the model being optimized and the iteration number (how many steps have been taken by the algorithm so far). You can click the Cancel button to abort the optimization process

When the optimization is successfully accomplished a number of spreadsheets and graphs are displayed in the following order:

Simplex

  1. Settings spreadsheet: Displays the setting of the Simplex algorithm as specified on the Simplex tab.
  2. Iterations spreadsheet - Displays
    1. predictions of the individual models for each iteration
    2. predictions of the ensemble (see Combine models option), together with its variance (for regression problems only) for each iteration.
  3. Results spreadsheet: Displays the final results of the algorithm including
    1. the set of independent values (i.e. Simplex solution) found by the Simplex algorithm together with the corresponding predictions of the individual models
    2. the set of independent values found by the Simplex algorithm together with the corresponding ensemble prediction (see the Combine models option) as well as its variance (for regression problems only).
  4. Iterations graph - Displays
    1. line plots of model predictions (y-axis) against Simplex iteration number (x-axis)
    2. line plot of ensemble predictions (should model combination be used, see the Combine models option), together with its errorbars (for regression problems only).

Grid and Random

  1. Settings spreadsheet: Displays the settings of the Grid and Random search algorithm as specified on the controls of the Grid and Random tabs.
  2. Results spreadsheet: Displays
    1. the set of independent values (together with the corresponding predictions) for which the models yielded a response closest to the desired value
    2. the set of independent values (together with the corresponding prediction) for which the ensemble yielded a response closest to the desired value. For regression problems the ensemble variance is also be displayed, indicating the level of agreement among the member models.
Element Name Description
Cancel Click this button to close the Response Optimization Startup Panel
Options For more information of descriptions of the commands on this menu, see also