Very simple reusable blocks for scikit-learn pipelines (inspired by scikit-lego)
Installation from the source tree:
python setup.py install
Or via pip from PyPI:
pip install scikit-duplo
The sci-kit duplo package contains multiple classes that you can use in a sci-kit
learn compatible pipeline. There are ensemble learning classes within the meta
subdirectory.
These classes expect you to pass in multiple other Sci-kit learn compatible
machine learning classes. It will use these to build an ensemble of models to
predict the target variable.
There are feature engineering classes inside the preprocessing
subdirectory. These are
ColumnTransformer compatible classes that expect to receive a dataframe and set of column
names that it will transform for the downstream pipeline processes.
LookupEncoder allows you to provide a custon dictionary of values for encoding categorical variables.