Search papers, labs, and topics across Lattice.
2
0
1
Pinpointing *why* your model's predictions are changing just got easier: ShapShift uses Shapley values and decision trees to explain prediction shifts by attributing them to changes in the conditional probabilities of interpretable data subgroups.
Popular local attribution methods like LIME and Shapley Values can assign importance to completely irrelevant features, even with independent features and exact computations.
