sciedm.aux_func.SurrogateData

sciedm.aux_func.SurrogateData(dataFrame=None, column=None, method='ebisuzaki', numSurrogates=10, alpha=None, smooth=0.8, outputFile=None)

Generate surrogate data

Three methods:

random_shuffle :

Sample the data with a uniform distribution.

ebisuzaki :

Journal of Climate. A Method to Estimate the Statistical Significance of a Correlation When the Data Are Serially Correlated. https://doi.org/10.1175/1520-0442(1997)010<2147:AMTETS>2.0.CO;2

Presumes data are serially correlated with low pass coherence. It is: “resampling in the frequency domain. This procedure will not preserve the distribution of values but rather the power spectrum (periodogram). The advantage of preserving the power spectrum is that resampled series retains the same autocorrelation as the original series.”

seasonal :

Presume a smoothing spline represents the seasonal trend. Each surrogate is a summation of the trend, resampled residuals, and possibly additive Gaussian noise. Default noise has a standard deviation that is the data range / 5.