contents.gifindex.gifprev1.gifnext1.gif

Sensitivity Tables

Another way to perform a sensitivity analysis is to generate a table of sensitivity values. Do this by pointing to Sensitivity Analysis in the Tools menu and clicking Sensitivity Table. Like the Sensitivity Graph option, this command brings up a dialog box where you specify your input and output nodes.

dph00128.gif

The result is a table similar to the following:

dph00129.gif

The first column in a sensitivity analysis table shows the percentage change on the output that is caused by a 1% change on the input. In the example above, a 1% change in Principal causes a 1% change in Payment, the output node. Similarly, a 1% change in Annual Rate causes a 0.76% change in Payment.

The second column in a sensitivity analysis table shows the absolute amount the output changes for a unit change in an input. For example, every dollar added to Principal causes Payment to go up by 0.73 cents. Every 100% added to Annual Rate adds $10,457 to Payment. And, finally, every year added to Term adds $8.83 to Payment.

You can use the absolute sensitivity values to create a linear equivalent to your model. The Payment model can be approximated using the equation

Payment = 1100.65 + 0.0073 Principal + 10457 Annual Rate - 8.8327 Years

The constant, 1100.65, is the current value of Payment in the original model.

You might have noticed in the sensitivity graph that Payment is not linearly related to Years. As Years increases the relative change in Payment goes down. The sensitivity table, however, does not show this effect. The sensitivity table presents only one value for the relationship between these two nodes. This value is the sensitivity at the input node's current value. More specifically, the sensitivity is calculated as the partial derivative of the output node as a function of the input node at the input node's current value.

Vanguard Software Applications

Our collaborative, web-based forecasting applications unite roles, teams and departments in achieving the most accurate forecasts in the world – bar none.