Update to sklearn 1.8

.github/workflows/deploy-gh-pages.yml

@@ -25,7 +25,7 @@ jobs:
     - name: Set up Python
       uses: actions/setup-python@v4
       with:
-        python-version: 3.9
+        python-version: 3.11
 
     - name: Install dependencies
       run: |

check_env.py

@@ -18,11 +18,11 @@
 pyversion_str = f"{sys.version_info.major}.{sys.version_info.minor}"
 pyversion = Version(pyversion_str)
 
-if pyversion < Version("3.8"):
+if pyversion < Version("3.11"):
     print(
         FAIL,
         (
-            "Python version 3.8 or above is required,"
+            "Python version 3.11 or above is required,"
             f" but {pyversion_str} is installed."
         ),
     )
@@ -66,7 +66,7 @@ def import_version(pkg, min_ver, fail_msg=""):
     "numpy": "1.16",
     "scipy": "1.2",
     "matplotlib": "3.0",
-    "sklearn": "1.6",
+    "sklearn": "1.8",
     "pandas": "1",
     "seaborn": "0.11",
     "notebook": "5.7",

environment-dev.yml

@@ -2,7 +2,8 @@ name: scikit-learn-course
 channels:
   - conda-forge
 dependencies:
-  - scikit-learn >= 1.6
+  - python=3.11
+  - scikit-learn >= 1.8
   - pandas >= 1
   - matplotlib-base
   - seaborn >= 0.13
@@ -13,4 +14,4 @@ dependencies:
   - packaging
   - pip
   - pip:
-    - jupyter-book >= 0.11
+    - jupyter-book < 2.0

environment.yml

@@ -4,7 +4,8 @@ channels:
   - conda-forge
 
 dependencies:
-  - scikit-learn >= 1.6
+  - python=3.11
+  - scikit-learn >= 1.8
   - pandas >= 1
   - matplotlib-base
   - seaborn >= 0.13

local-install-instructions.md

@@ -46,7 +46,7 @@ Using python in /home/lesteve/miniconda3/envs/scikit-learn-course
 [ OK ] numpy version 1.19.5
 [ OK ] scipy version 1.6.0
 [ OK ] matplotlib version 3.3.3
-[ OK ] sklearn version 1.6
+[ OK ] sklearn version 1.8
 [ OK ] pandas version 2.0
 [ OK ] seaborn version 0.13
 [ OK ] notebook version 6.2.0

notebooks/linear_models_ex_01.ipynb

@@ -43,7 +43,7 @@
     "penguins = pd.read_csv(\"../datasets/penguins_regression.csv\")\n",
     "feature_name = \"Flipper Length (mm)\"\n",
     "target_name = \"Body Mass (g)\"\n",
-    "data, target = penguins[[feature_name]], penguins[target_name]"
+    "data, target = penguins[[feature_name]], penguins[[target_name]]"
    ]
   },
   {

notebooks/linear_models_sol_01.ipynb

@@ -43,7 +43,7 @@
     "penguins = pd.read_csv(\"../datasets/penguins_regression.csv\")\n",
     "feature_name = \"Flipper Length (mm)\"\n",
     "target_name = \"Body Mass (g)\"\n",
-    "data, target = penguins[[feature_name]], penguins[target_name]"
+    "data, target = penguins[[feature_name]], penguins[[target_name]]"
    ]
   },
   {
@@ -152,7 +152,7 @@
     "def goodness_fit_measure(true_values, predictions):\n",
     "    # we compute the error between the true values and the predictions of our\n",
     "    # model\n",
-    "    errors = np.ravel(true_values) - np.ravel(predictions)\n",
+    "    errors = true_values - predictions\n",
     "    # We have several possible strategies to reduce all errors to a single value.\n",
     "    # Computing the mean error (sum divided by the number of element) might seem\n",
     "    # like a good solution. However, we have negative errors that will misleadingly\n",

python_scripts/linear_models_ex_01.py

@@ -40,7 +40,7 @@
 penguins = pd.read_csv("../datasets/penguins_regression.csv")
 feature_name = "Flipper Length (mm)"
 target_name = "Body Mass (g)"
-data, target = penguins[[feature_name]], penguins[target_name]
+data, target = penguins[[feature_name]], penguins[[target_name]]
 
 # %% [markdown]
 # ### Model definition

python_scripts/linear_models_sol_01.py

@@ -34,7 +34,7 @@
 penguins = pd.read_csv("../datasets/penguins_regression.csv")
 feature_name = "Flipper Length (mm)"
 target_name = "Body Mass (g)"
-data, target = penguins[[feature_name]], penguins[target_name]
+data, target = penguins[[feature_name]], penguins[[target_name]]
 
 # %% [markdown]
 # ### Model definition
@@ -106,7 +106,7 @@ def linear_model_flipper_mass(
 def goodness_fit_measure(true_values, predictions):
     # we compute the error between the true values and the predictions of our
     # model
-    errors = np.ravel(true_values) - np.ravel(predictions)
+    errors = true_values - predictions
     # We have several possible strategies to reduce all errors to a single value.
     # Computing the mean error (sum divided by the number of element) might seem
     # like a good solution. However, we have negative errors that will misleadingly

python_scripts/parameter_tuning_grid_search.py

@@ -91,7 +91,6 @@
     remainder="passthrough",
     # Silence a deprecation warning in scikit-learn v1.6 related to how the
     # ColumnTransformer stores an attribute that we do not use in this notebook
-    force_int_remainder_cols=False,
 )
 
 # %% [markdown]

python_scripts/parameter_tuning_nested.py

@@ -54,7 +54,6 @@
 preprocessor = make_column_transformer(
     (categorical_preprocessor, categorical_columns),
     remainder="passthrough",
-    force_int_remainder_cols=False,  # Silence a warning in scikit-learn v1.6.
 )
 
 # %%

python_scripts/parameter_tuning_randomized_search.py

@@ -73,7 +73,6 @@
 preprocessor = make_column_transformer(
     (categorical_preprocessor, categorical_columns),
     remainder="passthrough",
-    force_int_remainder_cols=False,  # Silence a warning in scikit-learn v1.6.
 )
 
 # %%

requirements-dev.txt

@@ -1,9 +1,9 @@
-scikit-learn>=1.6
+scikit-learn>=1.8
 pandas >= 1
 matplotlib
 seaborn >= 0.13
 plotly
-jupyter-book>=0.11
+jupyter-book < 2.0
 jupytext
 beautifulsoup4
 IPython

requirements.txt

@@ -1,4 +1,4 @@
-scikit-learn>=1.6
+scikit-learn>=1.8
 pandas >= 1
 matplotlib>=3.10
 seaborn >= 0.13