sigma_clip

astropy.stats.funcs.sigma_clip(data, sig=3, iters=1, cenfunc=<function median at 0x334e1b8>, varfunc=<function var at 0x321e230>, maout=False)

Perform sigma-clipping on the provided data.

This performs the sigma clipping algorithm - i.e. the data will be iterated over, each time rejecting points that are more than a specified number of standard deviations discrepant.

Note

scipy.stats.sigmaclip provides a subset of the functionality in this function.

Parameters :

data : array-like

The data to be sigma-clipped (any shape).

sig : float

The number of standard deviations (not variances) to use as the clipping limit.

iters : int or None

The number of iterations to perform clipping for, or None to clip until convergence is achieved (i.e. continue until the last iteration clips nothing).

cenfunc : callable

The technique to compute the center for the clipping. Must be a callable that takes in a 1D data array and outputs the central value. Defaults to the median.

varfunc : callable

The technique to compute the variance about the center. Must be a callable that takes in a 1D data array and outputs the width estimator that will be interpreted as a variance. Defaults to the variance.

maout : bool or ‘copy’

If True, a masked array will be returned. If the special string ‘inplace’, the masked array will contain the same array as data, otherwise the array data will be copied.

Returns :

filtereddata : numpy.ndarray or numpy.masked.MaskedArray

If maout is True, this is a masked array with a shape matching the input that is masked where the algorithm has rejected those values. Otherwise, a 1D array of values including only those that are not clipped.

mask : boolean array

Only present if maout is False. A boolean array with a shape matching the input data that is False for rejected values and True for all others.

Examples

This will generate random variates from a Gaussian distribution and return only the points that are within 2 sample standard deviation from the median:

>>> from astropy.tools.alg import sigma_clip
>>> from numpy.random import randn
>>> randvar = randn(10000)
>>> data,mask = sigma_clip(randvar, 2, 1)

This will clipping on a similar distribution, but for 3 sigma relative to the sample mean, will clip until converged, and produces a numpy.masked.MaskedArray:

>>> from astropy.tools.alg import sigma_clip
>>> from numpy.random import randn
>>> from numpy import mean
>>> randvar = randn(10000)
>>> maskedarr = sigma_clip(randvar, 3, None, mean, maout=True)