Module transforms 

Data transformations for statistical analysis.

Currently provides the Box-Cox power transformation, which maps non-normal positive data to approximate normality. The optimal transformation parameter λ is estimated via maximum likelihood.

References

Box, G. E. P. & Cox, D. R. (1964). “An analysis of transformations.” Journal of the Royal Statistical Society, Series B, 26(2), 211–252.

Enums

TransformError

Errors that can arise from Box-Cox transformations.

Functions

box_cox

Apply the Box-Cox power transformation to positive data.

estimate_lambda

Estimate the optimal Box-Cox λ via maximum likelihood.

inverse_box_cox

Invert a Box-Cox transformation.