[MRG] Fix Github action version v2 -> v4 and fix test with skorch update (#292)

* add contributor guide

* change v2 to v4

* try v3

* test fix tests

* fix tests

* fix tests?

Author: tgnassou

.github/workflows/testing.yml

@@ -26,11 +26,11 @@ jobs:
 
     # Install skada
     - name: Checking Out Repository
-      uses: actions/checkout@v2
+      uses: actions/checkout@v4
     # Cache 'skada' datasets
     - name: Create/Restore Skada Data Cache
       id: cache-skada_datasets
-      uses: actions/cache@v2
+      uses: actions/cache@v4
       with:
         path: ~/skada_datasets
         key: ${{ runner.os }}-v3
@@ -58,11 +58,11 @@ jobs:
 
     # Install skada
     - name: Checking Out Repository
-      uses: actions/checkout@v2
+      uses: actions/checkout@v4
     # Cache 'skada' datasets
     - name: Create/Restore Skada Data Cache
       id: cache-skada_datasets
-      uses: actions/cache@v2
+      uses: actions/cache@v4
       with:
         path: ~/skada_datasets
         key: ${{ runner.os }}-v3
@@ -100,11 +100,11 @@ jobs:
 
     # Install skada
     - name: Checking Out Repository
-      uses: actions/checkout@v2
+      uses: actions/checkout@v4
     # Cache 'skada' datasets
     - name: Create/Restore Skada Data Cache
       id: cache-skada_datasets
-      uses: actions/cache@v2
+      uses: actions/cache@v4
       with:
         path: ~/skada_datasets
         key: ${{ runner.os }}-v3

skada/metrics.py

@@ -10,11 +10,12 @@
 from copy import deepcopy
 
 import numpy as np
-from sklearn.base import BaseEstimator, clone
+from sklearn.base import clone
 from sklearn.linear_model import LogisticRegression
 from sklearn.metrics import balanced_accuracy_score, check_scoring
 from sklearn.model_selection import train_test_split
 from sklearn.neighbors import KernelDensity
+from sklearn.pipeline import Pipeline
 from sklearn.preprocessing import LabelEncoder, Normalizer
 from sklearn.utils import check_random_state
 from sklearn.utils.extmath import softmax
@@ -395,8 +396,7 @@ def _score(self, estimator, X, y, sample_domain=None, **kwargs):
             )
 
         has_transform_method = False
-
-        if not isinstance(estimator, BaseEstimator):
+        if not isinstance(estimator, Pipeline):
             # The estimator is a deep model
             if estimator.module_.layer_name is None:
                 raise ValueError("The layer_name of the estimator is not set.")