Separate transforms by groups by its meaning

[Mr-Geekman]

🚀 Feature Request

Separate transforms by groups by its meaning. It can help to make all of them usable.

Suggested groups:

1. Decomposition
   • change_point_trend.py
   • binseg.py
   • trend.py
   • detrend.py
   • stl.py
2. Feature Selection
   • filter.py
   • feature_selection.py
   • feature_importance.py
   • gale_shapley.py
3. Time
   • datetime_flags.py
   • fourier.py
   • holiday.py
   • lags.py
   • special_days.py
   • statistics.py
4. Math
   • add_constant.py
   • log.py
   • sklearn.py
   • scalers.py
   • power.py
5. Missing values
   • outliers.py
   • imputation.py
   • resample.py
6. Encoders
   • segment_encoder.py
7. NN
   • pytorch_forecasting.py

File base.py can be placed in the root of transforms/ folder.

Motivation

There are a lot of transforms and it can confuse a lot of users. If transform will be grouped by its meaning it will be easier to use the library.

Proposal

1. Separate transforms by the groups into folders.
2. Separate tests for transforms by the same groups.
3. Check that documentation shows these groups.

Test cases

No response

Alternatives

No response

Additional context

No response

Checklist

• I discussed this issue with ETNA Team