1 code implementation • 15 Nov 2022 • Naofumi Hama, Masayoshi Mase, Art B. Owen
Here we present some model-free methods that do not require access to the prediction function.
no code implementations • 31 May 2022 • Masayoshi Mase, Art B. Owen, Benjamin B. Seiler
The most popular methods for measuring importance of the variables in a black box prediction algorithm make use of synthetic inputs that combine predictor variables from multiple subjects.
1 code implementation • 25 May 2022 • Naofumi Hama, Masayoshi Mase, Art B. Owen
We find an expression for the expected value of the AUC under a random ordering of inputs to $f$ and propose an alternative area above a straight line for the regression setting.
Additive models Explainable Artificial Intelligence (XAI) +1
1 code implementation • 17 May 2021 • Benjamin B. Seiler, Masayoshi Mase, Art B. Owen
We use Shapley value to combine all of the reductions in log cardinality due to revealing a variable after some subset of the other variables has been revealed.
1 code implementation • 15 May 2021 • Masayoshi Mase, Art B. Owen, Benjamin B. Seiler
Cohort Shapley value is a model-free method of variable importance grounded in game theory that does not use any unobserved and potentially impossible feature combinations.
2 code implementations • 1 Nov 2019 • Masayoshi Mase, Art B. Owen, Benjamin Seiler
Instead of changing the value of a predictor we include or exclude subjects similar to the target subject on that predictor to form a similarity cohort.