About the Taguchi Robust Design Results Aggregate Parameter

When a Taguchi Robust Design component is created and configured, Isight creates an output aggregate parameter called Taguchi Results. After execution, the Taguchi component writes some basic results to this results aggregate depending on whether you are analyzing a static or dynamic system.

The following items are included in the Taguchi Results aggregate parameter:

  • Static Systems

    For static systems an Execution Summary is included in the Taguchi Results aggregate parameter. The Execution Summary provides a brief summary of the Taguchi Robust Design setup information (number of control and noise factors, number of signal levels, DOE techniques used, number of samples) and Taguchi response metric results in HTML format.

    In addition, the following items are included for each response:

    • SN Ratio. The Signal-to-noise ratio for each control experiment.
    • Mean. The mean response value for each control experiment.
    • Variance. The variance of all response values for each control experiment.
    • Loss. The value of Taguchi’s loss function for each control experiment.
  • Dynamic Systems

    For dynamic systems an Execution Summary is included in the Taguchi Results aggregate parameter. The Execution Summary provides a brief summary of the Taguchi Robust Design setup information (number of control and noise factors, number of signal levels, DOE techniques used, number of samples) and Taguchi response metric results in HTML format.

    In addition, the following items are included for each response:

    • SN Ratio. The dynamic signal-to-noise ratio for each control experiment.
    • Sensitivity. The sensitivity for each control experiment.
    • Beta. The slope of the signal/response relationship line for each control experiment.
    • St. The total sum of squares for each control experiment.
    • Sb. The variation caused by the linear effect.
    • Se. The variation associated with error and nonlinearity for each control experiment.
    • Ve. The error variance for each control experiment.