reset#

scikitplot.reset(_gallery_conf=None, _fname=None)[source]#

Reset global state for NumPy, scikit-learn, matplotlib, and seaborn.

This function restores common scientific computing and visualization libraries to their default configurations, preventing state leakage between scripts, examples, or plotting sessions.

It is especially useful in: - Documentation builds (e.g., with Sphinx-Gallery) - Interactive notebooks (Jupyter, Colab) - Streamlit or Dash apps (ensuring consistent reruns) - Automated testing pipelines - Educational or demo environments

What gets reset: - NumPy: Restores default error handling (np.seterr) and print formatting

(np.set_printoptions).

scikit-learn: Resets global configuration via sklearn.set_config, including options like assume_finite and working_memory.
matplotlib: Closes all open figures, resets rcParams, and reapplies the default style.
seaborn (if installed): Restores default theme and style to avoid residual settings.

Parameters:

_gallery_confdict, optional: Required for compatibility with Sphinx-Gallery’s reset_modules hook. Not used internally.
_fnamestr, optional: Filename of the current example (provided by Sphinx-Gallery). Not used internally.

See also

reset_numpy: Reset NumPy global state
reset_sklearn: Reset scikit-learn configuration
reset_matplotlib: Reset matplotlib figures and configuration
reset_seaborn: Reset seaborn theme (if available)

Notes

This function is idempotent and safe to call multiple times.

To use with Sphinx-Gallery, register it as:

>>> sphinx_gallery_conf = {
...     "reset_modules": "scikitplot.reset",
... }

You can also register it with atexit or use it in a pytest fixture:

>>> import atexit
>>> atexit.register(reset)

>>> import pytest
>>> @pytest.fixture(autouse=True)
... def clean_state():
...     yield
...     reset()

Examples

>>> from scikitplot._reset import reset
>>> reset()  # safely restore global state between examples

reset#

This Page