Using Average Values versus Simulation
This has to do with the idea of creating prediction models using average values. It is thought that if we use the average value for an uncertain input, we should get the average value for an output. While this might be true in some cases, it is not universally true for all models. This phenomenon is known as the Flaw of Averages.
Example: News vendor problem (situations where a one-time purchase decision must be made in the face of uncertain demand) - where a news vendor needs to make a decision about the number of newspapers to purchase to sell the next day.
If:
Order < Demand à Opportunity cost
Order > Demand à Leftover papers go to the recycling bin
Problem formulation:
Inputs for model à demand, selling price, cost and salvage value
Decision to be made for model à quantity to purchase
Output for model à total profit
Assumption for model à selling price > cost > salvage value
The execution is done using 2 models: Average model and Simulation model
The difference between the average model and the simulation model is that we have added uncertainty in the demand. The simulation model shows that the average value model is over-estimating the expected profit. This tells us that using an average value for an uncertain input, the demand in this case, does not necessarily result in the correct estimate for the average output value.
Through this example, we see how average values might result in misleading information, making the simulation model a better fit and alternative.
Example: News vendor problem (situations where a one-time purchase decision must be made in the face of uncertain demand) - where a news vendor needs to make a decision about the number of newspapers to purchase to sell the next day.
If:
Order < Demand à Opportunity cost
Order > Demand à Leftover papers go to the recycling bin
Problem formulation:
Inputs for model à demand, selling price, cost and salvage value
Decision to be made for model à quantity to purchase
Output for model à total profit
Assumption for model à selling price > cost > salvage value
The execution is done using 2 models: Average model and Simulation model
The difference between the average model and the simulation model is that we have added uncertainty in the demand. The simulation model shows that the average value model is over-estimating the expected profit. This tells us that using an average value for an uncertain input, the demand in this case, does not necessarily result in the correct estimate for the average output value.
Through this example, we see how average values might result in misleading information, making the simulation model a better fit and alternative.
