The first approach we considered was a link between attendance and foot traffic. Using the foot traffic data that we obtained, we calculated regression lines for attendance to Let’s Talk as the dependent variable to the foot traffic. The graphs are shown below with the equation displayed on the graph.
The R-squared values are as follows:
Based off of this simple analysis, we concluded that foot traffic was not the best angle from which to approach modeling attendance to a mental health program. We decided to turn to programs that were better suited for discrete event simulation.
In anyLogic, we planned to determine the number of students who will attend Let’s Talk by using the selectOutputIn and selectOutputOut blocks in anyLogic. These allow you to input a larger population and select individuals who would use the service from a uniform probability distribution. We planned to use the results of the attendance vs foot traffic graph to determine which individuals would use the resource. Once we selected the participants from a location that will attend Let’s Talk, we would be able to enter them into a system that uses queuing theory to determine average utilization, among other variables.
However, after researching the program more, we realized that anyLogic didn’t meet the needs for our project. It is a good tool for simulation visualization because it requires you to build the physical space that people move in and out of, such as a hospital ER. Our model, however, does not focus on the visual aspect of the simulation. Additionally, it is very difficult to generate individuals with specific characteristics, which we need to determine the number of people who will attend Let’s Talk.