TARNNResult.regime_weight

property TARNNResult.regime_weight: NDArray[float64]

1.0 above the threshold, else 0.0.

Strict on the upper side, matching the estimator’s z <= r assignment to the lower regime, so an observation landing exactly on the threshold is routed here the same way it was during training.

Example

>>> import numpy as np
>>> rng = np.random.default_rng(0)
>>> res = TARNN(rng.standard_normal(300), order=1).fit()
>>> int(res.regime_weight.sum()) == res.n_upper
True

Type:

Indicator of the upper regime