Analyzing Optimization Results

Once you have run an optimization, you can display history plots, download results, and export optimized designs.

The results of an optimization run are available while the Results Analytics session is active and are deleted when the session is closed. When you save an analytics case, only the approximation models created during the session are saved and the results of optimization runs over the approximation models are not saved.

This task shows you how to:

Before you begin: Run an optimization of your approximation model.
See Also
About Optimizations
Optimizing an Approximation Model

View the Parameter History Plots

You can manipulate the display of the history plots for each parameter.

  1. Run an optimization of your approximation model.
  2. Click Show improvements by score to switch the displayed history plots of the optimization parameters between all design points from the optimization run and only design points that score higher than the previously generated design points.

    All designs generated during an optimization run are scored using the objectives and constraints specified in the Define page. By default, Results Analytics only displays design points that score higher than the previously generated design points.

  3. Click Parameters to select which parameter plots are shown.

    If the number of optimization parameters is less than 10, the Optimize view displays all parameter plots. If there are more than 10 parameters, the Optimize view, by default, only displays parameters with an objective.

  4. Click Erase Optimization Results to erase the previously executed optimization results from the selected approximation model.

Download Optimization Results

You can download the results of your optimization for further analysis.

  1. Run an optimization of your approximation model.
  2. Click Download Optimization Results
    Results Analytics downloads a .TSV file containing the table of designs generated during the displayed optimization run along with their parameter values.

Add Designs from the Optimizations to the Analytics Case Data Set

You can export data points generated during an optimization run to the analytics case data set.

  1. Right-click on any point to display a context menu.
  2. Select to add data points to the analytics case data set. Select to add data points to both the analytics case data set and the analytics case basket.
  3. Return to the Explore page.
    The design appears in the data table with an optimization icon to indicate that it was generated by an optimization run. The design appears in the Data Sources filter with a data source name indicating the approximation model name and the optimization technique used. In the example below, Optimization_RSM_ASA is the data source name for a design from an optimization using the Adaptive Simulated Annealing (ASA) technique over a Response Surface Approximation Model (RSM).

    Note: Results Analytics scores your predicted design after generating the scores for the original data. As a result, the score for the generated design can be greater than 100 or less than 0.