Bayesian Decision Theory Blog 8

While surfing the interweb I came across this "normative approach" to decision making.

The Bayesian theory is pretty basic and is as follows:

1. Define a set of available actions
2. Define a set possible outcome of acts
3. Define a conditional probability distribution specifying the probability of each outcome given each available act.
4. Define a preference order ranking the possible outcomes distributions according to the desirability

The formulation of these definitions is really nothing more than a series of weighted averages that, in essence provide the user with the best decision possible.

This theory leaves presumes quite a bit. It presumes that there must be a defined amount of decision variable. In reality, there is not always a 0 to 100 percent probability of a course of action occurring. Sometimes the course of action is dependent upon other variable and are not up to chance.

Models are imperative to the decision making process, but oversimplification can be dangerous.