Explaining the root causes of unit-level changes
This work addresses the need for interpretable AI methods to explain unit-level changes, which is incremental as it builds on existing Shapley value and counterfactual approaches for specific attribution tasks.
The paper tackled the problem of explaining changes in output values for individual units by attributing them to changes in inputs and the underlying mechanism, proposing two counterfactual-based methods using Shapley values that satisfy key axioms. Through simulations and a case study on US earnings data, the methods demonstrated reliability and scalability, yielding sensible results in identifying drivers of change.
Existing methods of explainable AI and interpretable ML cannot explain change in the values of an output variable for a statistical unit in terms of the change in the input values and the change in the "mechanism" (the function transforming input to output). We propose two methods based on counterfactuals for explaining unit-level changes at various input granularities using the concept of Shapley values from game theory. These methods satisfy two key axioms desirable for any unit-level change attribution method. Through simulations, we study the reliability and the scalability of the proposed methods. We get sensible results from a case study on identifying the drivers of the change in the earnings for individuals in the US.