Computing the Shapley value
Same idea as isolating a member, but every comparison is now weighted by how often its group precedes that member in a random arrival order. The weighted sum is the Shapley value.
Putting it together
Shapley value = Σ weight × (payoff with − payoff without)
- weight
- share of random arrival orders in which this group is already there
- payoff
- bonus earned by the group
Marginal contribution across each pair of coalitions
Shapley value (weighted average)