Pushing the docs to dev/ for branch: main, commit ba1d23d13e402367c6401e07256867fdd5a4a0bf

Author: sklearn-ci

dev/_downloads/07fcc19ba03226cd3d83d4e40ec44385/auto_examples_python.zip

@@ -84,9 +84,10 @@
 set_config(transform_output="pandas")
 
 num_pipe = make_pipeline(SimpleImputer(), StandardScaler())
+num_cols = ["age", "fare"]
 ct = ColumnTransformer(
     (
-        ("numerical", num_pipe, ["age", "fare"]),
+        ("numerical", num_pipe, num_cols),
         (
             "categorical",
             OneHotEncoder(
@@ -114,3 +115,24 @@
 # This resets `transform_output` to its default value to avoid impacting other
 # examples when generating the scikit-learn documentation
 set_config(transform_output="default")
+
+# %%
+# When configuring the output type with :func:`config_context` the
+# configuration at the time when `transform` or `fit_transform` are
+# called is what counts. Setting these only when you construct or fit
+# the transformer has no effect.
+from sklearn import config_context
+
+scaler = StandardScaler()
+scaler.fit(X_train[num_cols])
+
+# %%
+with config_context(transform_output="pandas"):
+    # the output of transform will be a Pandas DataFrame
+    X_test_scaled = scaler.transform(X_test[num_cols])
+X_test_scaled.head()
+
+# %%
+# outside of the context manager, the output will be a NumPy array
+X_test_scaled = scaler.transform(X_test[num_cols])
+X_test_scaled[:5]

dev/_downloads/1b3f17ff0f112d5b77cbdb90f1c17046/plot_set_output.py

@@ -141,7 +141,7 @@
       },
       "outputs": [],
       "source": [
-        "from sklearn.compose import ColumnTransformer\nfrom sklearn.preprocessing import OneHotEncoder, StandardScaler\nfrom sklearn.impute import SimpleImputer\nfrom sklearn import set_config\n\nset_config(transform_output=\"pandas\")\n\nnum_pipe = make_pipeline(SimpleImputer(), StandardScaler())\nct = ColumnTransformer(\n    (\n        (\"numerical\", num_pipe, [\"age\", \"fare\"]),\n        (\n            \"categorical\",\n            OneHotEncoder(\n                sparse_output=False, drop=\"if_binary\", handle_unknown=\"ignore\"\n            ),\n            [\"embarked\", \"sex\", \"pclass\"],\n        ),\n    ),\n    verbose_feature_names_out=False,\n)\nclf = make_pipeline(ct, SelectPercentile(percentile=50), LogisticRegression())\nclf.fit(X_train, y_train)\nclf.score(X_test, y_test)"
+        "from sklearn.compose import ColumnTransformer\nfrom sklearn.preprocessing import OneHotEncoder, StandardScaler\nfrom sklearn.impute import SimpleImputer\nfrom sklearn import set_config\n\nset_config(transform_output=\"pandas\")\n\nnum_pipe = make_pipeline(SimpleImputer(), StandardScaler())\nnum_cols = [\"age\", \"fare\"]\nct = ColumnTransformer(\n    (\n        (\"numerical\", num_pipe, num_cols),\n        (\n            \"categorical\",\n            OneHotEncoder(\n                sparse_output=False, drop=\"if_binary\", handle_unknown=\"ignore\"\n            ),\n            [\"embarked\", \"sex\", \"pclass\"],\n        ),\n    ),\n    verbose_feature_names_out=False,\n)\nclf = make_pipeline(ct, SelectPercentile(percentile=50), LogisticRegression())\nclf.fit(X_train, y_train)\nclf.score(X_test, y_test)"
       ]
     },
     {
@@ -179,6 +179,53 @@
       "source": [
         "set_config(transform_output=\"default\")"
       ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "When configuring the output type with :func:`config_context` the\nconfiguration at the time when `transform` or `fit_transform` are\ncalled is what counts. Setting these only when you construct or fit\nthe transformer has no effect.\n\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "collapsed": false
+      },
+      "outputs": [],
+      "source": [
+        "from sklearn import config_context\n\nscaler = StandardScaler()\nscaler.fit(X_train[num_cols])"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "collapsed": false
+      },
+      "outputs": [],
+      "source": [
+        "with config_context(transform_output=\"pandas\"):\n    # the output of transform will be a Pandas DataFrame\n    X_test_scaled = scaler.transform(X_test[num_cols])\nX_test_scaled.head()"
+      ]
+    },
+    {
+      "cell_type": "markdown",
+      "metadata": {},
+      "source": [
+        "outside of the context manager, the output will be a NumPy array\n\n"
+      ]
+    },
+    {
+      "cell_type": "code",
+      "execution_count": null,
+      "metadata": {
+        "collapsed": false
+      },
+      "outputs": [],
+      "source": [
+        "X_test_scaled = scaler.transform(X_test[num_cols])\nX_test_scaled[:5]"
+      ]
     }
   ],
   "metadata": {