<< Click to Display Table of Contents >> Navigation:  Simulating > Sensitivity

To create a sensitivity run, expand the Analysis tab, right-click "Sensitivity", and select "Edit".

 

 

As with an optimization run, you'll need to provide a name (1), define the parameters (2), and adjust the settings (3) of the run.

 

To add a sensitivity run, click the icon. Once you've added the run, you can left-click on Name to set the name of the run.
 
Left-click <Params> field beneath Parameters to set the variables (beginning with Entity Type, which will populate all of the relevant variables in the list) and distribution for the sensitivity run. Currently, you can select between a normal, uniform, exponential, binomial, negative binomial, Weibull, Poisson, gamma, triangular, beta, or vector distribution.

 

Once you're finished setting the parameters for your sensitivity run, take a moment to adjust the settings (in particular the Max Samples setting, which determines how many iterations of runs will be made and defaults to 200) before running your model. There are several settings currently available for sensitivity runs with more in the works.

 

Univariate - one variable at a time.

Multivariate - all variables at the same time.

Latin Hypercube - ensures full range of each parameter is searched within the number of simulations (set via Max Samples) provided.

Grid - similar to multivariate with vector distributions for the parameters. The number of simulations specified is the number of divisions, and will be raised to the power determined by the number of search parameters to determine the actual number of simulations required. Grid runs can take a very long time and are meant to be stopped mid-run, but this feature is not fully developed. Crashes and other issues may occur, save your model before running!

 

Results will, by default, be displayed automatically during and at the end of the run.To adjust these settings, see Save Period below.

 

Graphs have two display types which can be toggled between in the upper-left: confidence bounds and individual runs. You can choose to isolate only those runs above or below a certain confidence bound in the confidence bound display setting by right-clicking on the section you're interested in and selecting the option from the drop-down list.

 

Save Lists

 

Save Lists are a useful tool for especially large models, or where you're only interested in the results of the run for a handful of variables. To access the save list, simply navigate to the Model Overview pane, right-click "save lists" and select "New Save List". This will automatically create a new Save List named "Save List1" under the save list tab. By either double-clicking or right-clicking and selecting "Edit", you'll see a display that includes every variable in your model broken down by entity, pictured below.

 

 

Selecting the check box for the entity "SensitivitySales" will select all of its variables (1). If you want to select individual variables in an entity, simply click the check box next to them instead (2).

 

You can also sort all of the variables by type (stock, flow, etc.) by clicking the relevant icons at the top of the save list (3).

 

Save Period

 

Similar to Save Lists, Save Period allows you to select a range of dates or times that are of particular interest to you and only save the results or changes that occur in the Sensitivity run within that time frame, or at certain time intervals. This will dramatically increase the speed and performance of the run, especially in larger Sensitivity runs. Changing the Save Period for a run will not effect the results of the run, it will simply change how much of the data is available to be seen and manipulated during and after the run has completed.

 

To adjust the Save Period, navigate to the Model Overview tab and right-click or double click on Run Control. On the far right of the newly opened Run Control there will be options for Save Time Step (1), Initial Save Time (2), Final Save Time (3), Initial Save Calender Time (4), and Final Save Calendar Time (5).

 

 

To set the Save Period you can either set a Save Time Step, which will determine how often Ventity saves and presents the information from the run to you, or the Save Times, with initial time in both cases relating to the time you've already set in Initial and Final times in your run control, or Calendar Times meaning the same but for Calendar time. Note that these settings will default to match those in the rest of your Run Control, meaning you'll save all of the data throughout the model run.

 

Finally, there are three additional options accessed through the Run Control for Sensitivity runs. These are: Sensitivity Results Path, Sensitivity Results Format, and Sensitivity Results in Memory.

 

 

Sensitivity Results in Memory is a toggle which determines whether or not the model will run and save the results of your sensitivity run in the memory of your computer or directly write the results to a file of the format you set in Sensitivity Results Format. Enabling this option will decrease the run time for your model, but will also not display any of the automatic results. If you want to see the results from your run, you'll have to load the data from the file you saved it as into the model.
 
Sensitivity Results Format determines the type of file that your sensitivity run results is saved as. Currently, the only options available are Tidy (.csv) and Binary (.vsrp).

 

Sensitivity Results Path is where the results of your sensitivity run will be saved. By default, this is in the "Results" folder of your model, but can be changed through this menu.