The problem I was working on here was to try to find the highest fidelity outcome I can while minimizing the computational time of a noisy model. The model takes two input parameters. As we increase the value of these parameters, the model fidelity increases. Unfortunately, the computational time goes up with the product of these two parameters. To see if there is a sweet spot, I run the model through a parameter sweep overnight using a variety of input parameter pairs. I am trying to determine if there is a setting for these two parameters that will give a best result while minimizing the computational time.
The first glance at the data is misleading at best. In the video we refine the view until the relevant information pops out in an obvious manner.