Optimization Problems

Overview

More than any other advanced method, the success of an optimization problem is directly a result of its setup. Essentially, all the real world factors and objectives must be turned into a parametrized “objective function” and “constraint functions.” The math that follows afterward is complex but standardized. Crafting these functions can turn into an art form, and direct experience with similar problem types and domains can be helpful. Sometimes trial and error is required, though ideally the first iterations of this can be done with business domain experts before launch.

Dynamic Approach

In certain cases, a solve-it-once approach is not enough. Rather, a dynamic approach is required where feedback is used to update beliefs and relative importances of objective components.