regressioninc.linear.robust module#

Functions to enable robust linear regression

regressioninc.linear.robust.mad(arr: ndarray, c: float = 0.6744897501960817, center: float | None = None) → float[source]#

The Median Absolute Deviation from the center

There is no immediately obvious median for complex data. Therefore, the absolute value is taken before calculating out the median.

Parameters:

  • arr (np.ndarray) – The data of which to take the

  • c (float, optional) – A scaling factor, by default Gaussian.ppf(3 / 4.0)

  • center (Optional[float], optional) – The center from which to calculate the median deviation, by default None. If it is None, the geometric median of the data will be used.

Returns:

mad = median(abs(a - center))/c

Return type:

float