Expand description
Propensity score processing, as DoubleML’s PSProcessor: optional
isotonic calibration of the cross-fitted propensity scores against the
treatment (or instrument), followed by clipping to [t, 1 − t].
IsotonicRegression is a port of scikit-learn’s
IsotonicRegression(out_of_bounds="clip", y_min=0, y_max=1).
Structs§
- Isotonic
Regression - Monotone (non-decreasing) regression of
yon a scalarx, predicting by linear interpolation and clipping inputs to the training range, as scikit-learn’sIsotonicRegression(out_of_bounds="clip").
Enums§
- Calibration
- Calibration of propensity scores before clipping.