This library is aimed to enhance data scientists' daily work efficiency.
Main features are listed below:
- dataframe_observer.py: exploratory data analysis
- eda_report.py: auto generate an EDA report for a dataset
- feature_handling.py: feature engineering for a dataset
- feature_handling2.py: feature engineering for training set and test set
- feature_generation.py: several feature generation methods
- feature_reduction.py: feature reduction by embedding methods
- feature_drift.py: visualize and detect feature drift between training set and test set
- drift_report.py: auto generate a feature drift report for 2 datasets
- modeling.py: fast modeling for binary/multiclass classification or regression tasks, including model validation & hyper-parameter searching
- evaluation.py: evaluate inference result in different aspects according to task type
- splitter_sampler.py: split or sample datasets
The following notebooks are quick tutorials for supervised learning:
- test_modeling_bin.ipynb: binary classification
- test_modeling_mlt.ipynb: multi-class classification
- test_modeling_reg.ipynb: regression
- Download the package
- [optional] Create a virtual env and activate it
- In terminal, run
$ cd ./Layzee
$ pip3 install . - To uninstall, run
$ pip3 uninstall layzee