statsmodels.robust.norms.Hampel

class statsmodels.robust.norms.Hampel(a=2.0, b=4.0, c=8.0)

Hampel function for M-estimation

Parameters:

a, b, cfloat, optional

The tuning constants for Hampel’s function. They have to be scalars that satisfy 0 < a <= b <= c, equal values are allowed. The default values are a, b, c = 2, 4, 8.

Methods

__call__(z)

Return the value of estimator rho applied to an input

Raises:

TypeError

If a, b or c is not a real scalar. Arrays are not accepted.

ValueError

If the constants do not satisfy 0 < a <= b <= c, or if one of them is nan.

See also

statsmodels.robust.norms.RobustNorm

Notes

Changed in version 0.15.1: The tuning constants are validated. Constants that are not positive or not ordered, and nan values, raise a ValueError instead of creating a degenerate norm. Array valued constants raise a TypeError, they never worked in the methods of the norm.

The tuning constants are validated when the norm is created and in _set_tuning_param. Plain attribute assignment after construction, for example norm.c = value, is not validated.

Methods

max_rho()
psi(z)	The psi function for Hampel's estimator
psi_deriv(z)	Derivative of psi function, second derivative of rho function
rho(z)	The robust criterion function for Hampel's estimator
weights(z)	Hampel weighting function for the IRLS algorithm

Properties

continuous
redescending