calc_AICc#

mufasa.UltraCube.calc_AICc(ucube, compID, mask, planemask=None, return_NSamp=True, expand=20)[source]#

Calculate the corrected Akaike Information Criterion (AICc) for a spectral cube model.

This function computes the AICc for a given model component by evaluating the residual sum of squares (RSS) and the effective sample size. The AICc values help compare the goodness-of-fit for models with varying complexities.

Parameters:

  • ucube (UltraCube) – An instance of the UltraCube class containing the spectral data, fitted models, and associated parameters.

  • compID (int or str) – The component ID or number of components in the model. If an integer, it is used to calculate the number of parameters as n_parameters = compID * 4.

  • mask (numpy.ndarray) – A 3D boolean array specifying which voxels (volume pixels) to include in the AICc calculation.

  • planemask (numpy.ndarray, optional) – A 2D spatial boolean array specifying which pixels to calculate AICc for. If provided, calculations are restricted to these pixels. Default is None.

  • return_NSamp (bool, optional) – If True, return the effective sample size along with the AICc values. Default is True.

  • expand (int, optional) – Number of spectral channels to expand the region where the RSS is calculated. Default is 20.

Returns:

  • AICc_map (numpy.ndarray) – A 2D array containing the computed AICc values for the specified regions.

  • NSamp_map (numpy.ndarray, optional) – A 2D array containing the effective sample size for each spatial pixel. Only returned if return_NSamp is True.

Notes

The corrected Akaike Information Criterion (AICc) is a modification of the AIC that accounts for finite sample sizes. It is given by:

AICc = AIC + (2k(k + 1)) / (N - k - 1)

where k is the number of parameters and N is the sample size.

Raises:

ValueError – If the input ucube does not contain valid models or residuals for the specified compID.

On this page

This Page