Parameter Sweep
Description
The mUQSA Parameter Sweep method is a simple toolkit for measuring uncertainty of the model for parameters taken uniformly from predefined ranges.
In simpler terms, think of it as a way to “sweep” through different scenarios and observe how changing one or more factors impacts your system. It’s like running multiple experiments without actually conducting them in a lab.
As an example, imagine you have a model that simulates a population’s response to a disease. You’re interested in how different disease durations, ranging from 4 days to 28 days, impact the outcome. With the Parameter Sweep method, you can explore this variation step by step. You may tell the Parameter Sweep method to consider values from 4 to 28 days, with 2-day increments. You can also systematically explore multiple parameters and reveal how different combinations of parameter values influence the system’s behavior. Thus, in a single analysis, we can investigate both disease duration and population size.
Features
The Parameter Sweep method is a simple and predictable approach for the elementary uncertainty quantification. Thanks to its intuitiveness, it allows getting quick insights into the model even for novice users.
However, the Parameter Sweep method does have certain limitations. It produces only basic statistics and can’t be used to perform Sensitivity Analysis. It is also most effective for systems that exhibit linear or moderate nonlinearity, rendering it less suitable for highly nonlinear systems that may necessitate more advanced methods. Moreover, the method assumes relatively simple probability distributions for parameters, posing challenges for systems with complex distributions. To obtain a comprehensive view of such models, all possible combinations of parameters may need to be evaluated, which is obviously highly inefficient.