Optimization Models

As we mentionned earlier, an Optimization model is based on a descriptive model where varibles that are under the modelers control are identified and simulated to find the best possible situation or outcome.

1. Sort the model's variables into 2 categories:

   1. Decision Variables: Variables you can change in the real world (i.e. order quantities, #people, budget allocations, etc)

   2. Statistical Assumptions: Variables you can't change in the reak world (i.e. interest rates, commodity prices, risk incidences, etc.)

2. Define a range of potential scenarios for each variable. Again 3 points works well in most situations: Lower Bound, Most Likely, Upper Bound

3. Depending on the tools available, you run multiple scenarios. Optimization could potetially be done with MS solver however, using tools such as Crystal Ball you can do a whole lot more damage because you account for the impact of variance...in other words the relative influence of variables on the final answer.)

4.  Make sure that the model makes sense and run your optimization as long as possible. If the numbers make Sense... run with that.

If you have any questions, please don't hesitate to contact me at etorkia@crystalballservices.com.

Eric