Simulation tab

The Simulation tab allows you to specify sensitivity analysis, refit probability distributions for simulated inputs and correlations between simulated inputs to new data, and modify the probability distribution associated with a simulated input.

The Simulated inputs grid contains an entry for each input field that is defined in the simulation plan. Each entry displays the name of the input and the probability distribution type associated with the input, along with a sample plot of the associated distribution curve. Each input also has an associated status icon (a colored circle with a check mark) that is useful when you are refitting distributions to new data. In addition, inputs may include a lock icon which indicates that the input is locked and cannot be modified or refit to new data in the Run Simulation dialog. To modify a locked input you will need to open the simulation plan in the Simulation Builder.

Each input is either simulated or fixed. Simulated inputs are those whose values are uncertain and will be generated by drawing from a specified probability distribution. Fixed inputs are those whose values are known and remain constant for each case generated in the simulation. To work with a particular input, select the entry for the input in the Simulated inputs grid.

Specifying sensitivity analysis

Sensitivity analysis allows you to investigate the effect of systematic changes in a fixed input or in a distribution parameter for a simulated input by generating an independent set of simulated cases—effectively, a separate simulation—for each specified value. To specify sensitivity analysis, select a fixed or simulated input and click Sensitivity Analysis. Sensitivity analysis is limited to a single fixed input or a single distribution parameter for a simulated input. See the topic Sensitivity Analysis (simulation) for more information.

Refitting distributions to new data

To automatically refit probability distributions for simulated inputs (and correlations between simulated inputs) to data in the active dataset:

  1. Verify that each of the model inputs is matched up with the correct field in the active dataset. Each simulated input is fit to the field in the active dataset specified in the Field dropdown list associated with that input. You can easily identify inputs that are unmatched by looking for inputs with a status icon that includes a check mark with a question mark, as shown below.
  2. Modify any necessary field matching by selecting Fit to a field in the dataset and selecting the field from the list.
  3. Click Fit All.

For each input that was fit, the distribution that most closely fits the data is displayed along with a plot of the distribution superimposed on a histogram (or bar chart) of the historical data. If an acceptable fit cannot be found then the Empirical distribution is used. For inputs that are fit to the Empirical distribution, you will only see a histogram of the historical data because the Empirical distribution is in fact represented by that histogram.

Note: For a complete list of status icons, see the topic Simulated Fields.

Modifying probability distributions

You can modify the probability distribution for a simulated input and optionally change a simulated input to a fixed input or vice versa.

  1. Select the input and select Manually set the distribution.
  2. Select the distribution type and specify the distribution parameters. To change a simulated input to a fixed input, select Fixed in the Type dropdown list.

Once you have entered the parameters for a distribution, the sample plot of the distribution (displayed in the entry for the input) will be updated to reflect your changes. For more information on manually specifying probability distributions, see the topic Simulated Fields.

Include user-missing values of categorical inputs when fitting. This specifies whether user-missing values of inputs with a Categorical distribution are treated as valid when you are refitting to data in the active dataset. System-missing values and user-missing values for all other types of inputs are always treated as invalid. All inputs must have valid values for a case to be included in distribution fitting and computation of correlations.