Results
By performing an uncertainty propagation (UP) analysis using the nonadaptive surrogate models, four tables are added to the output table group for the UP analysis.
•
A QoI Confidence interval table. It contains one row per QoI, with columns for the mean, standard deviation (STD), minimum, and maximum, followed by confidence intervals for 90%, 95%, and 99% likelihood.
•
A UP predicted QoI table. It contains the surrogate-model-predicted values for the QoIs for the Monte Carlo sampling points. This data table is used by KDE datasets that are also added by the analysis, one per QoI.
•
A UP predicted STD table. It contains the surrogate-model-predicted standard deviation for the Monte Carlo sampling points. This can be seen as the built-in surrogate model error estimation.
•
Finally, a Maximum entropy table. It contains the maximum relative standard deviation, one per global QoI.
When an adaptive surrogate model is used, an Adaptive maximum entropy table is also added to the output table group. It contains the maximum relative standard deviation, one per global QoI for all the adaption steps.
For each group of study-dependent QoIs, that is QoIs with the same parameter selections, there is an additional Maximum entropy and Adaptive maximum entropy table. Additionally, there is a Mean with 95% Prediction Interval table for each group. This table is derived from the QoI Confidence interval table and transposed with respect to the study-dependent parameter (time, frequency, …) for plotting. It contains the mean and the lower and upper bounds for the 95% prediction interval.
An Uncertainty Quantification plot group is also added with KDE plots, one for each QoI. For each group of study-dependent QoIs a table plot with error bars is added to visualize the dependency of the mean and 95% prediction intervals with respect to the study-dependent parameter.