flexml
Easy-to-use and flexible AutoML library for Python
APACHE-2.0 License
Ecosystems:
scikit-learn,
Python
Bot releases are hidden (Show)
v1.0.3
Improvements
- Set all logging_to_file params to False as default to prevent log file creation for the users for default #1ae7dbb
- Removed top_n_models param from supervised_base.py's class's start_experiment() func since there is actually no need for it when there is get_best_models() and get_model_by_name() functions #6bf77c4
- Connected requirements.txt file's library versions to setup.py to prevent double writing of the versions #7caaf9d
- Revised cv and n_jobs params' data types in the supervised_base.py and _model_tuner.py; Users were able to pass cv as None but in GridSearch and RandomizedSearch tuning algorithms, cv param as None is converted to integer 5 by default, to prevent this, set cv's default value as integer 3, and it can be minimum 2. Also, n_jobs's data type is changed to int only, instead of Optional[int] #37ccacc
- Improved tutorial notebooks and readme #70a0993 #17e1853 #593ecb2
- Removed test libraries from requirements.txt file to decrase the number of libraries to be installed during pip install. Users can install the test libraries by theirselves If they are going to make contribution and run the unit tests #0b3f98f
Error/Bug Fixes
- Sorting the new param grid with tuning option "quick" in _model_tuner.py was giving error when there is a None value in the param list since None and numeric values can't be compared, changed it by added None value catch exception to the sorting func #b50501c
Release on pip https://pypi.org/project/flexml/1.0.3/
flexml
- FlexML 1.0.2
Published by ozguraslank about 2 months ago
v1.0.2
Improvements
- improved experiment_size parameter's documentation in supervised base
Error/Bug Fixes
- Fixed parameter usage mistake in README file
Release on pip https://pypi.org/project/flexml/1.0.2/
Package Rankings
Top 34.55% on Pypi.org