**Model development - how to determine that you need Monte Carlo?**

As the world become more complicated our model must follow it. Now, there is impossible to calculate results directly. We must use numerical methods. Lets assume that our model become even more complicated, and that all input parameters are uncertain. In that case, uncertainty estimation methods must be used.

**Choosing input distributions**

On of the main problems in uncertainty propagation estimation is how to choose the input probability distribution. There are two ways:

- Calculate the parameters from the measurements
- Estimate the parameters on the current knowledge on the topic

**Way of sampling and statistics of output**

Sampling may be the crucial thing in choosing Monte Carlo as a uncertainty estimation method. Inadvertantly sampling will almost certain cost you in CPU time, which is most important when time consuming models are simulated.

**Statistics of the Monte Carlo output**

Main statistical parameters estimation are:

- mean value and
- variance/standard deviation