Optimizing an Approximation Model

You can run an optimization on an existing approximation model based on your objectives and constraints.

Before you begin: Define any objectives and constraints on your parameters in the Define page. You need at least one objective to start an optimization run. Create an approximation for your model from the Predict page. Your approximation model must be in a Ready state. For more information, see Creating and Editing Approximations.
See Also
About Optimizations
  1. From the Predict page, click Profiler to switch to the profiler view.
  2. For each input parameter, click and drag to position the profiler curve to the position of the value where the optimization will begin. Alternatively, enter the starting point in the ยต field.
  3. Click Optimize to switch to the optimization view.
  4. Click Settings , choose the optimization technique, and edit the options.

    The options displayed are specific to the optimization technique. For more information on the available optimization techniques, see About the Optimization Techniques. For information on configuring the optimization techniques, see Optimization Configuration.

  5. Optional: Select the lock in the Parameters - Input section to keep one or more input parameters constant when performing the optimization run.

    Results Analytics keeps the value of these input parameters constant at your specified starting value, while varying all the other input parameters during the optimization run.

  6. Click Start Optimization .

    Tip: To end the optimization run before it is complete, click Stop Optimization .

    The optimization status displays information about your approximation.

    Status Description
    Ready The optimization run is complete and successful.
    Running The optimization is executing.
    None No optimization results are available for the selected approximation model.
    Aborted The optimization run has been stopped before completion.
    Out of date The displayed optimization run results are out of date. Optimization results are out of date if the underlying approximation model is out of date or if you modify the objectives and constraints.

    History plots appear, displaying the progress of the optimization run for each parameter. You can resize, reorder, and filter each plot. For more information on viewing an optimization, see Analyzing Optimization Results.