gemmr.util.pc_spectrum_decay_constant

gemmr.util.pc_spectrum_decay_constant(X=None, pc_spectrum=None, expl_var_ratios=(0.3, 0.9), plot=False)

Estimate powerlaw decay constant.

Parameters:

  • X (np.ndarray (n_samples, n_features)) – the principal component spectrum will be estimated for data matrix X. Either X or pc_spectrum must be given

  • pc_spectrum (np.ndarray (n_pcs,)) – directly the principal component spectrum. Either X or pc_spectrum must be given

  • expl_var_ratios (tuple of floats) – decay constants will be estimated by fitting a linear regression to the principal component spectrum using the number of components that explain this amount of variance. If multiple values for expl_var_ratios are given, multiple decay constants will be returned

  • plot (bool) – if True, a plot of the spectrum and illustrating the decay constant will be returned in addition to the decay constant

Returns:

  • decay_constant (np.ndarray (len(expl_var_ratios),)) – estimated powerlaw decay constant

  • panel (hv.Overlay) – if argument plot == True, a panel showing the principal component spectrum and illustrating the decay constant