Refactoring preprocessing

[valer1435]

Solving issue https://github.com/nccr-itmo/FEDOT/issues/812

The main changes:

1. Ohe-hot-encoding and simple_imputation are deleted from default model pool
2. Cleaning preprocessor become after fold evaluation
3. Due to (3) we need to fit pipeline to calculate intermediate metric

[nicl-nno]

Общий вопрос - а сейчас PreprocessingBuilder ведь по прежнему будет добавить one_hot и прочее в начальные приближения. Или нет?

[codecov[bot]]

Codecov Report

Merging #816 (4c880d9) into master (74823a9) will increase coverage by 0.08%. The diff coverage is 89.83%.

@@            Coverage Diff             @@
##           master     #816      +/-   ##
==========================================
+ Coverage   87.37%   87.45%   +0.08%     
==========================================
  Files         184      184              
  Lines       12904    12997      +93     
==========================================
+ Hits        11275    11367      +92     
- Misses       1629     1630       +1     

Impacted Files	Coverage Δ
...api/api_utils/assumptions/preprocessing_builder.py	100.00% <ø> (ø)
...implementations/data_operations/sklearn_filters.py	89.92% <57.14%> (-2.52%)	:arrow_down:
...t/api/api_utils/assumptions/assumptions_handler.py	85.10% <100.00%> (+0.66%)	:arrow_up:
fedot/core/caching/pipelines_cache.py	85.71% <100.00%> (-1.13%)	:arrow_down:
fedot/core/caching/preprocessing_cache.py	89.28% <100.00%> (-4.05%)	:arrow_down:
...t/core/optimisers/objective/data_objective_eval.py	100.00% <100.00%> (ø)
fedot/core/pipelines/pipeline.py	98.24% <100.00%> (+0.01%)	:arrow_up:
fedot/core/validation/split.py	100.00% <100.00%> (ø)
fedot/preprocessing/preprocessing.py	99.31% <100.00%> (ø)
fedot/core/data/data_preprocessing.py	93.15% <0.00%> (-4.11%)	:arrow_down:
... and 17 more

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

[valer1435]

Общий вопрос - а сейчас PreprocessingBuilder ведь по прежнему будет добавить one_hot и прочее в начальные приближения. Или нет?

Добавляеет. Удалю. В остальном - не знаю, имеет ли смысл удалять все упоминания об one-hot и imputation. Все таки скоро будем перерабатывать и вернемся к этому

[nicl-nno]

Все упоминания безусловно нет, достаточно чтобы в эволюции такие операции не фигурировали.

[valer1435]

Переработал немного использование кэша. Удалось полностью его вынести из pipeline.

[pep8speaks]

Hello @valer1435! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:

There are currently no PEP 8 issues detected in this Pull Request. Cheers! :beers:

Comment last updated at 2022-08-19 08:58:15 UTC