search

UniprJRC / FSDA

Flexible Statistics and Data Analysis (FSDA) extends MATLAB for a robust analysis of data sets affected by different sources of heterogeneity. It is open source software licensed under the European Union Public Licence (EUPL). FSDA is a joint project by the University of Parma and the Joint Research Centre of the European Commission.
https://uniprjrc.github.io/FSDA/
European Union Public License 1.1
83 stars 38 forks source link

e.g. regression #38

Open DanGitR opened 2 years ago

DanGitR commented 2 years ago

I'd suggest expanding Powertra.m to implement standardization (to mean=0 and std. dev. =1) after BoxCox or Yeo-Johnson is applied.

This would make Powertra.m equivalent to the scikit-learn PowerTransformer frequently used in Python DA workflows.

UniprJRC commented 2 years ago

Dear Daniel, Thank you for your suggestion. Option 'standardize' has been added to Powertra.m. The associated HTML documentation is at

http://rosa.unipr.it/FSDA/Powertra.html

An example which implements this option is at http://rosa.unipr.it/FSDA/Powertra.html#ExtraExample_3

The modifications are already in the github repo of FSDA and they all be present inside the FSDA toolbox starting from release 8.5.26 Please, let us know if your suggestion has been taken into account. For any other suggestion, please do not hesitate to contact us Marco (on behalf of all the FSDA team)