Step 1. Statistical preprocessor: sample generation
The first step
in the sample generation phase is to select ranges and distributions
(probability distribution functions, pdf) for the input factors.
This selection makes use of the best information available on the
statistical properties of the input factors. In some instances, it is
possible to get empirical estimates of pdf’s from available
underlying data for the input factors. The effort put at this stage is
related to the scope of the analysis: if the analysis is at an
exploratory stage, then rather crude pdf’s could be adequate:
the cost of getting better pdf’s may be relatively high.
factors setup
contains all the APIs necessary to manage factors (create, delete, save, etc...)
The second step is to select a sampling method among the available ones. Note the choice of the sampling method has implications on what type of sensitivity analysis the user is able to perform later on (for example, if we use FAST sampling, we cannot perform the analysis using the method of Morris).
sampling methods
contains all the APIs relevant for the definition of the sampling method
If the input factors are not independent, they experience some kind of correlation that has to be declared in order obtain reliable results from the analysis.
correlation
contains the commands to set and remove correlation between input factors
The last phase is the actual generation of the sample from the input pdf’s.
sample generation
contains the commands for sample generation, save and load data