linear_shrink_target

hd.linear_shrink_target(cov, target, step=0.05, max_iter=100)

Linearly shrink a covariance matrix until a condition number target is reached. Useful for reducing the impact of outliers in nerive().

Parameters

covnumpy.ndarray, shape = (p, p)

The covariance matrix.

target: float > 1

The highest acceptable condition number.

stepfloat > 0

The linear shrinkage parameter for each step.

max_iterint > 1

The maximum number of iterations until giving up.

Returns

covnumpy.ndarray, shape = (p, p)

The linearly shrunk covariance matrix estimate.