

autogoal.contrib.sklearn

Note

This module contains wrappers for several estimators and transformers from scikit-learn.

Warning

Importing this module requires sklearn with a version equal or greater than 0.22. You can either install it manually or with pip install autogoal[sklearn].

Most of the classes and functions inside this module deal with the automatic generation of wrappers and thus are considered private API.

The main public functionality exposed by this module is the function find_classes, which allows to enumerate the wrappers implemented in this module applying some filters.

Note

You can manually import any wrapper class directly from autogoal.contrib.sklearn._generated buy beware that namespace changes wildly from version to version and classes in it might disappear or change their signature anytime.

try:
    import sklearn

    major, minor, *rest = sklearn.__version__.split(".")
    assert int(major) == 0 and int(minor) >= 22
except:
    print("(!) Code in `autogoal.contrib.sklearn` requires `sklearn=^0.22`.")
    print("(!) You can install it with `pip install autogoal[sklearn]`.")
    raise ImportError()

Filter out sklearn convergence and joblib warnings

import warnings
from sklearn.exceptions import ConvergenceWarning

warnings.filterwarnings("ignore", message=".*resource_tracker.*")
warnings.filterwarnings("ignore", category=ConvergenceWarning)
warnings.filterwarnings("ignore", module="sklearn")
warnings.filterwarnings("ignore", module="scipy")
warnings.filterwarnings("ignore", module="numpy")


from ._generated import *
from ._manual import *