Main Content

Ensemble Statistics

Calculate ensemble statistics when you want to characterize ensemble behavior as a whole. Also calculate ensemble statistics when you plan to generate residual signals for feature extraction. Unlike ensemble signals that contain multiple members, the resulting statistics signals each contain only one member. If you want to simply view the ensemble statistics without generating any new variables, use the Ensemble Summary plot in the plot gallery instead.

  • Ensemble Statistics — Select which statistics you want to generate. For each statistic that you select, the value is calculated at each time point across all members. The result is a single-member signal for that statistic. For example, if you select Ensemble mean, the single computed value at each time point is the mean of the values of all ensemble members at that time point.

  • Interpolate data — Select this option to provide a common time base when the different members in the ensemble have different sampling points.

    • Interpolation method — Select the interpolation method to use. For information on interpolation methods, see the Interpolation Method section of interp1.

    • Sampling frequency — The default auto setting calculates the mean sampling frequency across all members.

  • Group by — Select a condition variable to group by when you want to separate the statistics corresponding to different groups. For example, if your condition variable indicates healthy or faulty, you can separate the healthy statistics from the faulty statistics.

  • Plot results — Select this option if you want to plot all statistical results in the same plot. If you want to choose which results you plot, use Signal Trace with the resulting statistics signals.

An alternative method of generating ensemble statistics is by showing signal variation in a signal trace plot of your ensemble signal. That plotting option generates identical statistics to using Ensemble Statistics.

The software stores the results of the computation in the SummaryData dataset. The new variable names include the source signal name with the suffix mean, min. max, or std.