Running a sensitivity analysis

Having designed and tested the simulation, the analysts share the simulation plan with the finance department, who will run the simulation under different scenarios in order to determine the feasibility of the new distribution center. The finance department is confident that the estimated startup costs of $1,000,000 are reasonably accurate, but there is uncertainty in the growth rate of revenue and insufficient information about the growth rate to treat it as a simulated input. Using sensitivity analysis, the finance department can run the simulation over a chosen set of values of the growth rate, generating a separate distribution of the net present value for each value of the growth rate.

  1. From the menus choose:

    Analyze > Simulation...

  2. Select Open an Existing Simulation Plan in the Simulation: Model Source dialog and click Continue.
  3. In the Open a Simulation Plan dialog, browse to where you saved the simulation plan file and open the file.
    Figure 1. Run Simulation: Simulation tab
    Run Simulation: Simulation tab

    This opens the plan file in the Run Simulation dialog, which provides a simpler interface than the Simulation Builder and is designed for running simulations once a simulation plan has been created. Notice that the input growth, on which we'll base the sensitivity analysis, is already selected in the Simulated inputs list.

  4. Click Sensitivity Analysis....
    Figure 2. Sensitivity Analysis dialog
    Sensitivity Analysis dialog
  5. Select Iterate.
  6. Click the Value column for the second row in the Parameter value by iteration grid, enter 10 and then press Enter.
  7. Enter 12 in the third row of the Value column and then click Continue.
    Figure 3. Run Simulation: Simulation tab
    Run Simulation: Simulation tab

    The Parameters grid for the fixed input growth shows the three specified values of 8, 10 and 12. This represents a range of values from the original conservative estimate of 8% to a more aggressive estimate of 12% for the growth rate of revenue.

    Each value of the fixed input growth results in a separate set of simulated data based on using that particular value in the predictive model. Each such set of simulated data is referred to as an iteration and results in a separate distribution function of the target. This allows you to examine the effect on the distribution of the target due to varying a fixed input.

    Although not done for this example, you can also vary the value of a distribution parameter for a simulated input and investigate the effect on the distribution of the target. For example, you might consider varying the mode of the triangular distribution for operating costs. Note, however, that you are limited to varying either a single fixed input or a single distribution parameter for a simulated input.

  8. Select the Output tab.
    Figure 4. Run Simulation: Output tab
    Run Simulation: Output tab
  9. In the Density Functions group, select Cumulative Distribution Function (CDF) and deselect Probability Density Function (PDF).

    The probability density function and the cumulative distribution function provide the same information from different viewpoints. When iterations are present, the density functions from the separate iterations are overlaid on a single chart and will often show better separation when viewed as cumulative functions.

  10. Deselect Quartiles table in the Other Charts and Tables group.
  11. Click Run.

Next