aimclub / FEDOT

Automated modeling and machine learning framework FEDOT
https://fedot.readthedocs.io
BSD 3-Clause "New" or "Revised" License
627 stars 86 forks source link

Fedot.fit runs tuning even if with_tuing=False in composer_params #361

Closed J3FALL closed 2 years ago

J3FALL commented 3 years ago

In Fedot constructor there is a parameter - _composerparams and it is a dictionary with some external params, like learning_time, cv_folds, etc. However, if with_tuning is set explicitly, it is still running tuning after composing.

image

Example of code how it can be reproduced:

def forecast_by_fedot(forecast_length, train_input, predict_input):
    task_parameters = TsForecastingParams(forecast_length=forecast_length)
    model = Fedot(problem='ts_forecasting', task_params=task_parameters,
                  composer_params={'with_tuning': False})
    pipeline = model.fit(features=train_input)
    pipeline.show()
    forecast = model.predict(features=predict_input)

    return forecast

P.S. I've running the latest version from master branch

nicl-nno commented 3 years ago

Перевел на @v1docq , т.к. это можно пофиксить при обновлении API.

v1docq commented 2 years ago

Нашел ошибку в коде, на этапе выбора пресета флаг для тюнера переназначался вне зависимости от флага который мы подавали на входе. Исправил.