skmap.io.process.SircleTransformer#

class SircleTransformer(wv_0, wv_f=[], wv_p=[], wm_0=None, wm_f=[], wm_p=[], use_mask=False, return_den=False, keep_original_values=True, S=[], backend='dense', n_jobs=4, verbose=False)[source]#

Bases: Transformer

Parameters:

  • data – N_timeseries x N_samples matrix where the time series are stored one per each row

  • w_0 – convolution coefficent associated with the present

  • w_f – convolution coefficents associated with the future

  • w_p – convolution coefficents associated with the past

  • use_mask (bool) – decide if to use a mask for weights renormalization

  • return_den (bool) – in case of usage of the mask will return the denominator matrix in the Hadamard division

  • S – optional N_timeseries x N_samples matrix where per element scalings are stored

  • use_fft_backend – force usage of FFT backend computation of the convolution

  • n_jobs (int) – number of CPU to be used in parallel

Methods

run