Solvers
Ledoit-Wolf — Data-Estimated Shrinkage
Data-driven covariance shrinkage estimator.
How It Works
Shrinks covariance toward scaled identity with data-estimated intensity (Ledoit & Wolf, 2004). Faithful NumPy port of scikit-learn reference.
Parameters
| Parameter | Default | Description |
|---|---|---|
| shrinkage | "ledoit_wolf" | Data-estimated intensity (ignores shrinkage_alpha) |
from vyntri import Vyntri
from vyntri.data import split
s = split("./my_dataset", train=0.7, test=0.2, seed=42)
model = Vyntri(shrinkage="ledoit_wolf")
model.fit(train=s.train, val=s.val)Note: Ledoit-Wolf is not supported by the continual learning path. Use update() with shrinkage="diagonal".