Sensitivity Analysis
Sensitivity Analysis is defined as the study of how the variation in the output of a model can be attributed to variations in the inputs. Often, sensitivity analyses are carried out to determine which input variables have the most significant influence on the models outcome. This type of analysis is typically used to identify potential areas for improvement or to indicate areas where changes should be made to the model.
Sensitivity analyses are typically used to assess predictions by examining the effect that changing a particular input has on the models outcome. For example, in a risk analysis, sensitivity analysis might be used to evaluate the potential effects of environmental changes on the future of a business. In this case, a sensitivity analysis would be conducted by varying one input (e.g. the price of oil) and then examining the effects of this change on the models output (e.g. revenue).
In most applications, sensitivity analyses are conducted by varying a single input while keeping the remaining inputs constant. This approach is called one-way sensitivity analysis and can provide insights into how sensitive the model is to changes in the input. However, it is also possible to use multiple-way sensitivity analyses to understand the effects of varying multiple inputs simultaneously.
When attempting to interpret the results of a sensitivity analysis, the following considerations should be taken into account:
1. The magnitude of the change in output must be compared to the magnitude of the change in the input. If the output is relatively small given the input change, then the model may not be particularly sensitive. If, however, the output undergoes a significant change even when only a small change is made to the input, then the model is likely to be highly sensitive.
2. The direction of the change must be taken into account. If the output changes in the opposite direction to that of the input then this suggests that the two are related in some way and that changing the input parameters can cause significant changes in the output.
3. The scale of the input variables must be considered. If the input variables are on a different scale, then this may affect the correlation between the input and output.
Sensitivity analysis is a powerful tool for exploring the implications of model parameters and for better understanding the ways in which inputs and outputs interact. It can be used to inform decisions and to help identify weaknesses in models. By conducting sensitivity analyses, researchers can gain insights into the behaviour of systems, and can identify areas for improvement.
