Decision Optimization

Purpose

Choose which actions a team executes within limited hours, comparing the value-per-hour rule, the exact optimum and the relaxation bound.

Variables

• v_ivalue of action i
• a_ieffort of action i, in hours
• bhours available
• x_i1 if action i is executed

Assumptions

• Twelve synthetic actions; values and efforts come from the seed.
• Efforts are multiples of 0.25 h, so the exact optimum is found by dynamic programming.
• Actions are independent of one another.

Model

Data this would need in practice

• Value estimates per action, with their uncertainty.
• Measured effort per action.
• Real capacity and policy constraints.

Interpretation

Always optimum ≤ bound. The gap between the rule and the optimum is the room for a better method; the gap between optimum and bound is intrinsic to actions that cannot be split.

Limitations

• Values depend on forecasts that may be wrong.
• Interactions between actions are ignored.
• Policy limits must be added as constraints.