A regular problem in multiple regression is asserting the relative influence of the predictors in the model. Net Effects is a well known technique that are used to measure the shares that each predictor have on the target variable in the coefficient of multiple determination R^2. In the case of correlated inputs, net effects fail to give interpretable results which calls for other positive interpretable metrics.

Incremental Net Effect is estimated as the marginal influence of the predictor in all possible coalation of predictors on the target variable. This idea directly relates to Shapley values in cooperative game theory.