Commit
This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository.
DAAL-based RandomForestRegressor is known to give non-deterministic r…
…esults for the same inputs Consequentially, new in sklearn 0.21 check_fit_idempotent test for all estimators fail for daal4py.sklearn.ensemble.RandomForestRegressor The failure is small: ``` Mismatch: 15% Max absolute difference: 5.9604645e-08 Max relative difference: 3.3125482e-07 x: array([ 0.316434, 0.428736, -0.397625, 0.521435, 0.150143, 0.119337, -0.55916 , -0.38844 , 0.509067, -2.236972, 0.015745, 0.138834, -0.053688, -0.949258, -0.595061, -0.044984, 0.319812, -1.284615, -1.002518, 0.529771], dtype=float32) y: array([ 0.316434, 0.428736, -0.397625, 0.521435, 0.150143, 0.119337, -0.55916 , -0.38844 , 0.509067, -2.236972, 0.015745, 0.138834, -0.053688, -0.949258, -0.595061, -0.044984, 0.319812, -1.284615, -1.002518, 0.529771], dtype=float32) ``` and goes away if comparison is slightly relaxed. Intel(R) DAAL team is working on a fix. For now, `check_fit_idempotent` is skipped for RandomForestRegressor.
- Loading branch information