Useful Applications of Simulation Modeling

Useful Applications of Simulation Modeling

Useful Applications of Simulation Modeling

By November 30, 2015 Booz Allen, Data Science No Comments

Useful Applications of Simulation Modeling

Simulation Modeling is a structured approach to discovering key variable relationships within a system. Systems take on many forms across sectors, from agriculture to aerospace and defense to zoology. These systems are generally finite and operate within a set of defined business rules, often forcing decision makers to make difficult tradeoffs that can result in a range of profitable, or costly, outcomes.

To clarify options and outcomes, a simulation model is a powerful tool, aiding the decision maker in identifying whether and to what extent Key Performance Indicators (KPIs) can be achieved. This is accomplished by strategically varying model inputs — such as resource levels, funding amounts, policy measures, and business rules — and measuring the resulting outputs predicted by the model. By designing a series of varying model input parameters, the decision maker will get an idea of the range of possible system outcomes, and how specific changes to inputs will change the outcomes. This is useful in many scenarios, including making hiring decisions, formulating a budget, or forecasting operational metrics.

Consider the following example on how using a simulation model could be helpful to an HR manager who needs to develop a staffing plan.

Challenge: Top Hospital in the Washington, DC area has finalized next year’s staffing budget and identified how many surgeons it would like to hire from its residency program. How many applicants must the HR manager consider for its residency program this year in order to have the sufficient number of resident graduates to hire next year?

Background: When hiring full time surgeons, Top Hospital prefers to hire exclusively from its own residency program as it leads to shorter acclimation times and higher retention rates. The hospital’s residency program is a one-year cohort that requires a series of certifications, clinics, and board examinations to successfully graduate. At each knowledge check point, the resident is given two attempts to pass. After the second failed attempt, the resident is removed from the program entirely.

Unfortunately for the HR manager, the cohort graduation rate is highly variable and dependent on many factors. The most useful indicators of residency program success tend to be undergraduate and medical school GPA, undergraduate and graduate institution, undergraduate major, medical school specialty, and previous fellowship type.

Approach: The HR manager should design a simulation model that projects the success rate of each of the current applicants to determine if the projected graduation class is sufficient to meet the surgeon staffing plan.

This is accomplished by labeling each applicant with their individual success indicator attributes, and then programming the simulation with the likelihood of the combination of those attributes passing or failing each certification point, along with historical offer acceptance rates. When the simulation is run, it will provide an output of how many residents are expected to successfully complete the entire program.

Conclusion: The HR manager at Top Hospital is able to determine that it has enough applicants with the right mix of attributes to successfully graduate and hire the desired number of surgeons. The simulation model can then be used to hone in on a list of target schools revealed by the model to have the highest success rates in completing the residency program.

As seen in this use case and many others, Simulation Modeling can be a helpful in making inferences about a set of conditions that lead to a favorable or unfavorable outcome. Because behavior within a simulation model is represented with informed, random distributions, it’s important to remember that no simulation output provides an absolute solution. There are always caveats and sensitivities a user must digest and combine with system knowledge to make the best decision for an organization.

—Written by Tim Flynn