3. Simulate

With software such as Excel, the simulation phase is very straightforward, especially if neither correlated variables nor empirical laws are used. In the latter two cases, you need to code in VBA to obtain simulated values that respect these properties. In the simplest cases, when using probability laws known to Excel (or easily programmable, see Step "Understanding the GUM S1 method"), simulation is immediately accessible. Obviously, it is necessary to know the model of interest (which may require VBA programming if it is an algorithm) and the parameters of the input data (probability laws and their parameters, i.e. mean and standard deviation or lower and upper bounds of variations depending on the type of law).

As mentioned above, we simulate possible values for each input x _i and then calculate the resultant y via...