fixes max_eval_time_mins and max_eval_time_mins to allow None and inf - also allows steady state estimator to use template search spaces

setup.py

@@ -27,15 +27,15 @@ def calculate_version():
 
 ''',
     zip_safe=True,
-    install_requires=['numpy==1.26.4',
+    install_requires=['numpy>=1.26.4',
                       'scipy>=1.3.1',
-                      'scikit-learn>=1.4.2,<1.6',
+                      'scikit-learn>=1.6',
                       'update_checker>=0.16',
                       'tqdm>=4.36.1',
                       'stopit>=1.1.1',
                       'pandas>=2.2.0',
                       'joblib>=1.1.1',
-                      'xgboost>=1.7.0',
+                      'xgboost>=3.0.0',
                       'matplotlib>=3.6.2',
                       'traitlets>=5.8.0',
                       'lightgbm>=3.3.3',

tpot/builtin_modules/arithmetictransformer.py

@@ -35,12 +35,12 @@
 """
 import random
 import numpy as np
-from sklearn.base import BaseEstimator, TransformerMixin
+from sklearn.base import TransformerMixin, BaseEstimator 
 
 
 #operations are done along axis
 #TODO potentially we could do operations on every combo (mul would be all possible pairs multiplied with each other)
-class ArithmeticTransformer(BaseEstimator,TransformerMixin):
+class ArithmeticTransformer(TransformerMixin, BaseEstimator):
 
     #functions = ["add", "mul_neg_1", "mul", "safe_reciprocal", "eq","ne","ge","gt","le","lt", "min","max","0","1"]
     def __init__(self, function,):
@@ -140,7 +140,7 @@ def issorted(x, rev=False):
 
 
 
-class AddTransformer(BaseEstimator,TransformerMixin):
+class AddTransformer(TransformerMixin, BaseEstimator):
     def __init__(self):
           """
           A transformer that adds all elements along axis 1.
@@ -163,7 +163,7 @@ def transform_helper(self, X):
             X = np.expand_dims(X,0)
         return np.expand_dims(np.sum(X,1),1)
 
-class mul_neg_1_Transformer(BaseEstimator,TransformerMixin):
+class mul_neg_1_Transformer(TransformerMixin, BaseEstimator):
     def __init__(self):
         """
         A transformer that multiplies all elements by -1.
@@ -186,7 +186,7 @@ def transform_helper(self, X):
             X = np.expand_dims(X,0)
         return X*-1
 
-class MulTransformer(BaseEstimator,TransformerMixin):
+class MulTransformer(TransformerMixin, BaseEstimator):
 
     def __init__(self):
         """
@@ -210,7 +210,7 @@ def transform_helper(self, X):
             X = np.expand_dims(X,0)
         return np.expand_dims(np.prod(X,1),1)
 
-class SafeReciprocalTransformer(BaseEstimator,TransformerMixin):
+class SafeReciprocalTransformer(TransformerMixin, BaseEstimator):
 
     def __init__(self):
         """
@@ -234,7 +234,7 @@ def transform_helper(self, X):
             X = np.expand_dims(X,0)
         return np.divide(1.0, X.astype(float), out=np.zeros_like(X).astype(float), where=X!=0) #TODO remove astypefloat?
 
-class EQTransformer(BaseEstimator,TransformerMixin):
+class EQTransformer(TransformerMixin, BaseEstimator):
 
     def __init__(self):
         """
@@ -258,7 +258,7 @@ def transform_helper(self, X):
             X = np.expand_dims(X,0)
         return np.expand_dims(np.all(X == X[0,:], axis = 1),1).astype(float)
 
-class NETransformer(BaseEstimator,TransformerMixin):
+class NETransformer(TransformerMixin, BaseEstimator):
 
     def __init__(self):
         """
@@ -284,7 +284,7 @@ def transform_helper(self, X):
 
 
 
-class GETransformer(BaseEstimator,TransformerMixin):
+class GETransformer(TransformerMixin, BaseEstimator):
 
     def __init__(self):
         """
@@ -310,7 +310,7 @@ def transform_helper(self, X):
         return  result.astype(float)
 
 
-class GTTransformer(BaseEstimator,TransformerMixin):
+class GTTransformer(TransformerMixin, BaseEstimator):
     def __init__(self):
           """
           A transformer that takes checks if all elements in a row are greater than 0.
@@ -335,7 +335,7 @@ def transform_helper(self, X):
         return  result.astype(float)
 
 
-class LETransformer(BaseEstimator,TransformerMixin):
+class LETransformer(TransformerMixin, BaseEstimator):
     def __init__(self):
         """
         A transformer that takes checks if all elements in a row are less than or equal to 0.
@@ -360,7 +360,7 @@ def transform_helper(self, X):
         return  result.astype(float)
 
 
-class LTTransformer(BaseEstimator,TransformerMixin):
+class LTTransformer(TransformerMixin, BaseEstimator):
     def __init__(self):
         """
         A transformer that takes checks if all elements in a row are less than 0.
@@ -385,7 +385,7 @@ def transform_helper(self, X):
         return  result.astype(float)
 
 
-class MinTransformer(BaseEstimator,TransformerMixin):
+class MinTransformer(TransformerMixin, BaseEstimator):
     def __init__(self):
         """
         A transformer that takes the minimum of all elements in a row.
@@ -410,7 +410,7 @@ def transform_helper(self, X):
 
 
 
-class MaxTransformer(BaseEstimator,TransformerMixin):
+class MaxTransformer(TransformerMixin, BaseEstimator):
 
     def __init__(self):
           """
@@ -435,7 +435,7 @@ def transform_helper(self, X):
         return np.expand_dims(np.amax(X,1),1)
 
 
-class ZeroTransformer(BaseEstimator,TransformerMixin):
+class ZeroTransformer(TransformerMixin, BaseEstimator):
 
     def __init__(self):
           """
@@ -460,7 +460,7 @@ def transform_helper(self, X):
         return np.zeros((X.shape[0],1))
 
 
-class OneTransformer(BaseEstimator,TransformerMixin):
+class OneTransformer(TransformerMixin, BaseEstimator):
     def __init__(self):
           """
           A transformer that returns an array of ones.
@@ -484,7 +484,7 @@ def transform_helper(self, X):
         return np.ones((X.shape[0],1))
 
 
-class NTransformer(BaseEstimator,TransformerMixin):
+class NTransformer(TransformerMixin, BaseEstimator):
 
     def __init__(self, n):
         """

tpot/builtin_modules/column_one_hot_encoder.py

@@ -36,7 +36,7 @@
 import numpy as np
 from scipy import sparse
 
-from sklearn.base import BaseEstimator, TransformerMixin
+from sklearn.base import TransformerMixin, BaseEstimator 
 from sklearn.utils import check_array
 from sklearn.preprocessing import OneHotEncoder, OrdinalEncoder
 import sklearn
@@ -69,7 +69,7 @@ def _X_selected(X, selected):
 
 
 
-class ColumnOneHotEncoder(BaseEstimator, TransformerMixin):
+class ColumnOneHotEncoder(TransformerMixin, BaseEstimator ):
 
 
     def __init__(self, columns='auto', drop=None, handle_unknown='infrequent_if_exist', sparse_output=False, min_frequency=None,max_categories=None):
@@ -205,7 +205,7 @@ def transform(self, X):
             else:
                 return np.hstack((X_not_sel, X_sel))
 
-class ColumnOrdinalEncoder(BaseEstimator, TransformerMixin):
+class ColumnOrdinalEncoder(TransformerMixin, BaseEstimator ):
 
 
     def __init__(self, columns='auto', handle_unknown='error', unknown_value = -1, encoded_missing_value = np.nan, min_frequency=None,max_categories=None):

tpot/builtin_modules/estimatortransformer.py

@@ -34,14 +34,14 @@
 
 """
 from numpy import ndarray
-from sklearn.base import BaseEstimator, TransformerMixin
+from sklearn.base import TransformerMixin, BaseEstimator 
 from sklearn.model_selection import cross_val_predict
 from sklearn.utils.validation import check_is_fitted
 from sklearn.utils.metaestimators import available_if
 import numpy as np
 from sklearn.utils.validation import check_is_fitted
 
-class EstimatorTransformer(BaseEstimator, TransformerMixin):
+class EstimatorTransformer(TransformerMixin, BaseEstimator ):
     def __init__(self, estimator, method='auto', passthrough=False, cross_val_predict_cv=None):
         """
         A class for using a sklearn estimator as a transformer. When calling fit_transform, this class returns the out put of cross_val_predict

tpot/builtin_modules/feature_encoding_frequency_selector.py

@@ -6,7 +6,7 @@
 from sklearn.base import BaseEstimator
 from sklearn.feature_selection._base import SelectorMixin
 
-class FeatureEncodingFrequencySelector(BaseEstimator, SelectorMixin):
+class FeatureEncodingFrequencySelector(SelectorMixin, BaseEstimator):
     """Feature selector based on Encoding Frequency. Encoding frequency is the frequency of each unique element(0/1/2/3) present in a feature set. 
      Features are selected on the basis of a threshold assigned for encoding frequency. If frequency of any unique element is less than or equal to threshold, the feature is removed.  """
 

tpot/builtin_modules/feature_set_selector.py

@@ -45,7 +45,7 @@
 
 
 #TODO clean this up and make sure it works
-class FeatureSetSelector(BaseEstimator, SelectorMixin):
+class FeatureSetSelector(SelectorMixin, BaseEstimator):
     """
     Select predefined feature subsets.
 
@@ -109,8 +109,10 @@ def fit(self, X, y=None):
 
     # def transform(self, X):
 
-    def _get_tags(self):
-        tags = {"allow_nan": True, "requires_y": False}
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        tags.input_tags.allow_nan = True
+        tags.target_tags.required = False # formally requires_y
         return tags
 
     def _get_support_mask(self):

tpot/builtin_modules/feature_transformers.py

@@ -38,14 +38,14 @@
 
 
 import numpy as np
-from sklearn.base import BaseEstimator, TransformerMixin
+from sklearn.base import TransformerMixin, BaseEstimator 
 from sklearn.utils import check_array
 from sklearn.decomposition import PCA
 
 from .one_hot_encoder import OneHotEncoder, auto_select_categorical_features, _X_selected
 
 
-class CategoricalSelector(BaseEstimator, TransformerMixin):
+class CategoricalSelector(TransformerMixin, BaseEstimator ):
     """Meta-transformer for selecting categorical features and transform them using OneHotEncoder.
 
     Parameters
@@ -101,7 +101,7 @@ def transform(self, X):
             return ohe.fit_transform(X_sel)
 
 
-class ContinuousSelector(BaseEstimator, TransformerMixin):
+class ContinuousSelector(TransformerMixin, BaseEstimator ):
     """Meta-transformer for selecting continuous features and transform them using PCA.
 
     Parameters

tpot/builtin_modules/genetic_encoders.py

@@ -5,11 +5,11 @@
 """
 
 import numpy as np
-from sklearn.base import BaseEstimator, TransformerMixin
+from sklearn.base import TransformerMixin, BaseEstimator 
 from sklearn.utils import check_array
 
 
-class DominantEncoder(BaseEstimator, TransformerMixin):
+class DominantEncoder(TransformerMixin, BaseEstimator ):
     """This class contains the function definition for encoding the input features as a Dominant genetic model.
     The encoding used is AA(0)->1, Aa(1)->1, aa(2)->0. """
 
@@ -47,7 +47,7 @@ def transform(self, X, y=None):
 
         return X_transformed
 
-class RecessiveEncoder(BaseEstimator, TransformerMixin):
+class RecessiveEncoder(TransformerMixin, BaseEstimator ):
     """This class contains the function definition for encoding the input features as a Recessive genetic model.
     The encoding used is AA(0)->0, Aa(1)->1, aa(2)->1. """
 
@@ -85,7 +85,7 @@ def transform(self, X, y=None):
 
         return X_transformed
 
-class HeterosisEncoder(BaseEstimator, TransformerMixin):
+class HeterosisEncoder(TransformerMixin, BaseEstimator ):
     """This class contains the function definition for encoding the input features as a Heterozygote Advantage genetic model.
     The encoding used is AA(0)->0, Aa(1)->1, aa(2)->0. """
 
@@ -123,7 +123,7 @@ def transform(self, X, y=None):
 
         return X_transformed
 
-class UnderDominanceEncoder(BaseEstimator, TransformerMixin):
+class UnderDominanceEncoder(TransformerMixin, BaseEstimator ):
     """This class contains the function definition for encoding the input features as a Under Dominance genetic model.
     The encoding used is AA(0)->2, Aa(1)->0, aa(2)->1. """
 
@@ -162,7 +162,7 @@ def transform(self, X, y=None):
         return X_transformed
 
 
-class OverDominanceEncoder(BaseEstimator, TransformerMixin):
+class OverDominanceEncoder(TransformerMixin, BaseEstimator ):
     """This class contains the function definition for encoding the input features as a Over Dominance genetic model.
     The encoding used is AA(0)->1, Aa(1)->2, aa(2)->0. """
 

tpot/builtin_modules/imputer.py

@@ -39,7 +39,7 @@
 import numpy as np
 from scipy import sparse
 
-from sklearn.base import BaseEstimator, TransformerMixin
+from sklearn.base import TransformerMixin, BaseEstimator 
 from sklearn.utils import check_array
 from sklearn.preprocessing import OneHotEncoder
 import sklearn
@@ -50,7 +50,7 @@
 import sklearn.compose
 
 
-class ColumnSimpleImputer(BaseEstimator, TransformerMixin):
+class ColumnSimpleImputer(TransformerMixin, BaseEstimator ):
     def __init__(self,  columns="all",         
                         missing_values=np.nan,
                         strategy="mean",

tpot/builtin_modules/nn.py

@@ -109,7 +109,7 @@ def set_params(self, **parameters):
         return self
 
 
-class PytorchClassifier(PytorchEstimator, ClassifierMixin):
+class PytorchClassifier(ClassifierMixin, PytorchEstimator):
     @abstractmethod
     def _init_model(self, X, y): # pragma: no cover
         pass
@@ -131,7 +131,6 @@ def fit(self, X, y):
         self
             Fitted estimator.
         """
-        # pylint: disable=no-member
 
         self._init_model(X, y)
 
@@ -193,7 +192,6 @@ def validate_inputs(self, X, y):
         return (X, y)
 
     def predict(self, X):
-        # pylint: disable=no-member
 
         X = check_array(X, accept_sparse=True)
         check_is_fitted(self, 'is_fitted_')
@@ -295,8 +293,11 @@ def _init_model(self, X, y):
         self.train_dset_len = len(train_dset)
         self.device = device
 
-    def _more_tags(self):
-        return {'non_deterministic': True, 'binary_only': True}
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        tags.non_deterministic = True
+        tags.target_tags.single_output = True
+        return tags
 
 class PytorchMLPClassifier(PytorchClassifier):
     """Multilayer Perceptron, implemented in PyTorch, for use with TPOT.
@@ -347,6 +348,9 @@ def _init_model(self, X, y):
         )
         self.train_dset_len = len(train_dset)
         self.device = device
-
-    def _more_tags(self):
-        return {'non_deterministic': True, 'binary_only': True}
+
+    def __sklearn_tags__(self):
+        tags = super().__sklearn_tags__()
+        tags.non_deterministic = True
+        tags.target_tags.single_output = True
+        return tags