|
| 1 | +# flake8: noqa |
| 2 | +""" |
| 3 | +======================================= |
| 4 | +Release Highlights for scikit-learn 1.2 |
| 5 | +======================================= |
| 6 | +
|
| 7 | +.. currentmodule:: sklearn |
| 8 | +
|
| 9 | +We are pleased to announce the release of scikit-learn 1.2! Many bug fixes |
| 10 | +and improvements were added, as well as some new key features. We detail |
| 11 | +below a few of the major features of this release. **For an exhaustive list of |
| 12 | +all the changes**, please refer to the :ref:`release notes <changes_1_2>`. |
| 13 | +
|
| 14 | +To install the latest version (with pip):: |
| 15 | +
|
| 16 | + pip install --upgrade scikit-learn |
| 17 | +
|
| 18 | +or with conda:: |
| 19 | +
|
| 20 | + conda install -c conda-forge scikit-learn |
| 21 | +
|
| 22 | +""" |
| 23 | + |
| 24 | +# %% |
| 25 | +# Pandas output with `set_output` API |
| 26 | +# ----------------------------------- |
| 27 | +# scikit-learn's transformers now support pandas output with the `set_output` API. |
| 28 | +# To learn more about the `set_output` API see the example: |
| 29 | +# :ref:`sphx_glr_auto_examples_miscellaneous_plot_set_output.py` and |
| 30 | +# # this `video, pandas DataFrame output for scikit-learn transformers |
| 31 | +# (some examples) <https://youtu.be/5bCg8VfX2x8>`__. |
| 32 | + |
| 33 | +import numpy as np |
| 34 | +from sklearn.datasets import load_iris |
| 35 | +from sklearn.preprocessing import StandardScaler, KBinsDiscretizer |
| 36 | +from sklearn.compose import ColumnTransformer |
| 37 | + |
| 38 | +X, y = load_iris(as_frame=True, return_X_y=True) |
| 39 | +sepal_cols = ["sepal length (cm)", "sepal width (cm)"] |
| 40 | +petal_cols = ["petal length (cm)", "petal width (cm)"] |
| 41 | + |
| 42 | +preprocessor = ColumnTransformer( |
| 43 | + [ |
| 44 | + ("scaler", StandardScaler(), sepal_cols), |
| 45 | + ("kbin", KBinsDiscretizer(encode="ordinal"), petal_cols), |
| 46 | + ], |
| 47 | + verbose_feature_names_out=False, |
| 48 | +).set_output(transform="pandas") |
| 49 | + |
| 50 | +X_out = preprocessor.fit_transform(X) |
| 51 | +X_out.sample(n=5, random_state=0) |
| 52 | + |
| 53 | +# %% |
| 54 | +# Interaction constraints in Histogram-based Gradient Boosting Trees |
| 55 | +# ------------------------------------------------------------------ |
| 56 | +# :class:`~ensemble.HistGradientBoostingRegressor` and |
| 57 | +# :class:`~ensemble.HistGradientBoostingClassifier` now supports interaction constraints |
| 58 | +# with the `interaction_cst` parameter. For details, see the |
| 59 | +# :ref:`User Guide <interaction_cst_hgbt>`. In the following example, features are not |
| 60 | +# allowed to interact. |
| 61 | +from sklearn.datasets import load_diabetes |
| 62 | +from sklearn.ensemble import HistGradientBoostingRegressor |
| 63 | + |
| 64 | +X, y = load_diabetes(return_X_y=True, as_frame=True) |
| 65 | + |
| 66 | +hist_no_interact = HistGradientBoostingRegressor( |
| 67 | + interaction_cst=[[i] for i in range(X.shape[1])], random_state=0 |
| 68 | +) |
| 69 | +hist_no_interact.fit(X, y) |
| 70 | + |
| 71 | +# %% |
| 72 | +# Experimental Array API support in :class:`~discriminant_analysis.LinearDiscriminantAnalysis` |
| 73 | +# -------------------------------------------------------------------------------------------- |
| 74 | +# Experimental support for the `Array API <https://data-apis.org/array-api/latest/>`_ |
| 75 | +# specification was added to :class:`~discriminant_analysis.LinearDiscriminantAnalysis`. |
| 76 | +# The estimator can now run on any Array API compliant libraries such as |
| 77 | +# `CuPy <https://docs.cupy.dev/en/stable/overview.html>`__, a GPU-accelerated array |
| 78 | +# library. For details, see the :ref:`User Guide <array_api>`. |
0 commit comments