snowflake-ml-python 1.1.2__py3-none-any.whl → 1.2.1__py3-none-any.whl

snowflake/ml/modeling/linear_model/tweedie_regressor.py

@@ -54,6 +54,18 @@ _PROJECT = "ModelDevelopment"
 _SUBPROJECT = "".join([s.capitalize() for s in "sklearn.linear_model".replace("sklearn.", "").split("_")])
 
 
+def _is_fit_predict_method_enabled() -> Callable[[Any], bool]:
+    def check(self: BaseTransformer) -> TypeGuard[Callable[..., object]]:
+        return False and callable(getattr(self._sklearn_object, "fit_predict", None))
+    return check
+
+
+def _is_fit_transform_method_enabled() -> Callable[[Any], bool]:
+    def check(self: BaseTransformer) -> TypeGuard[Callable[..., object]]:
+        return False and callable(getattr(self._sklearn_object, "fit_transform", None))
+    return check
+
+
 class TweedieRegressor(BaseTransformer):
     r"""Generalized Linear Model with a Tweedie distribution
     For more details on this class, see [sklearn.linear_model.TweedieRegressor]
@@ -206,7 +218,9 @@ class TweedieRegressor(BaseTransformer):
         self.set_label_cols(label_cols)
         self.set_passthrough_cols(passthrough_cols)
         self.set_drop_input_cols(drop_input_cols)
-        self.set_sample_weight_col(sample_weight_col)        
+        self.set_sample_weight_col(sample_weight_col)
+        self._use_external_memory_version = False
+        self._batch_size = -1        
         deps: Set[str] = set([f'numpy=={np.__version__}', f'scikit-learn=={sklearn.__version__}', f'cloudpickle=={cp.__version__}'])
         
         self._deps = list(deps)
@@ -290,11 +304,6 @@ class TweedieRegressor(BaseTransformer):
         if isinstance(dataset, DataFrame):
             session = dataset._session
             assert session is not None  # keep mypy happy
-            # Validate that key package version in user workspace are supported in snowflake conda channel
-            # If customer doesn't have package in conda channel, replace the ones have the closest versions
-            self._deps = pkg_version_utils.get_valid_pkg_versions_supported_in_snowflake_conda_channel(
-                pkg_versions=self._get_dependencies(), session=session, subproject=_SUBPROJECT)
-            
             # Specify input columns so column pruning will be enforced
             selected_cols = self._get_active_columns()
             if len(selected_cols) > 0:
@@ -322,7 +331,9 @@ class TweedieRegressor(BaseTransformer):
             label_cols=self.label_cols,
             sample_weight_col=self.sample_weight_col,
             autogenerated=self._autogenerated,
-            subproject=_SUBPROJECT
+            subproject=_SUBPROJECT,
+            use_external_memory_version=self._use_external_memory_version,
+            batch_size=self._batch_size,
         )
         self._sklearn_object = model_trainer.train()
         self._is_fitted = True
@@ -593,6 +604,22 @@ class TweedieRegressor(BaseTransformer):
                 # each row containing a list of values.
                 expected_dtype = "ARRAY"
 
+            # If we were unable to assign a type to this transform in the factory, infer the type here.
+            if expected_dtype == "":
+                # If this is a clustering transformer, if the number of output columns does not equal the number of clusters the response will be an "ARRAY"
+                if hasattr(self._sklearn_object, "n_clusters") and getattr(self._sklearn_object, "n_clusters") != len(self.output_cols):
+                    expected_dtype = "ARRAY"
+                # If this is a decomposition transformer, if the number of output columns does not equal the number of components the response will be an "ARRAY"
+                elif hasattr(self._sklearn_object, "n_components") and getattr(self._sklearn_object, "n_components") != len(self.output_cols):
+                    expected_dtype = "ARRAY"
+                else:
+                    output_types = [signature.as_snowpark_type() for signature in _infer_signature(dataset[self.input_cols], "output", use_snowflake_identifiers=True)]
+                    # We can only infer the output types from the input types if the following two statemetns are true:
+                    # 1) All of the output types are the same. Otherwise, we still have to fall back to variant because `_sklearn_inference` only accepts one type.
+                    # 2) The length of the input columns equals the length of the output columns. Otherwise the transform will likely result in an `ARRAY`.
+                    if all(x == output_types[0] for x in output_types) and len(output_types) == len(self.output_cols):
+                        expected_dtype = convert_sp_to_sf_type(output_types[0])
+            
             output_df = self._batch_inference(
                 dataset=dataset,
                 inference_method="transform",
@@ -608,8 +635,8 @@ class TweedieRegressor(BaseTransformer):
 
         return output_df
     
-    @available_if(original_estimator_has_callable("fit_predict"))  # type: ignore[misc]
-    def fit_predict(self, dataset: Union[DataFrame, pd.DataFrame]) -> npt.NDArray[Any]:
+    @available_if(_is_fit_predict_method_enabled())  # type: ignore[misc]
+    def fit_predict(self, dataset: Union[DataFrame, pd.DataFrame]) -> Union[Any, npt.NDArray[Any]]:
         """ Method not supported for this class.
 
 
@@ -622,13 +649,21 @@ class TweedieRegressor(BaseTransformer):
         Returns:
             Predicted dataset.
         """
-        if False:
-            self.fit(dataset)
-            assert self._sklearn_object is not None
-            labels : npt.NDArray[Any] = self._sklearn_object.labels_
-            return labels
-        else:
-            raise NotImplementedError
+        self.fit(dataset)
+        assert self._sklearn_object is not None
+        return self._sklearn_object.labels_
+
+
+    @available_if(_is_fit_transform_method_enabled())  # type: ignore[misc]
+    def fit_transform(self, dataset: Union[DataFrame, pd.DataFrame]) -> Union[Any, npt.NDArray[Any]]:
+        """ 
+        Returns:
+            Transformed dataset.
+        """
+        self.fit(dataset)
+        assert self._sklearn_object is not None
+        return self._sklearn_object.embedding_
+
 
     def _get_output_column_names(self, output_cols_prefix: str, output_cols: Optional[List[str]] = None) -> List[str]:
         """ Returns the list of output columns for predict_proba(), decision_function(), etc.. functions.

snowflake/ml/modeling/manifold/isomap.py

@@ -54,6 +54,18 @@ _PROJECT = "ModelDevelopment"
 _SUBPROJECT = "".join([s.capitalize() for s in "sklearn.manifold".replace("sklearn.", "").split("_")])
 
 
+def _is_fit_predict_method_enabled() -> Callable[[Any], bool]:
+    def check(self: BaseTransformer) -> TypeGuard[Callable[..., object]]:
+        return False and callable(getattr(self._sklearn_object, "fit_predict", None))
+    return check
+
+
+def _is_fit_transform_method_enabled() -> Callable[[Any], bool]:
+    def check(self: BaseTransformer) -> TypeGuard[Callable[..., object]]:
+        return True and callable(getattr(self._sklearn_object, "fit_transform", None))
+    return check
+
+
 class Isomap(BaseTransformer):
     r"""Isomap Embedding
     For more details on this class, see [sklearn.manifold.Isomap]
@@ -199,7 +211,9 @@ class Isomap(BaseTransformer):
         self.set_label_cols(label_cols)
         self.set_passthrough_cols(passthrough_cols)
         self.set_drop_input_cols(drop_input_cols)
-        self.set_sample_weight_col(sample_weight_col)        
+        self.set_sample_weight_col(sample_weight_col)
+        self._use_external_memory_version = False
+        self._batch_size = -1        
         deps: Set[str] = set([f'numpy=={np.__version__}', f'scikit-learn=={sklearn.__version__}', f'cloudpickle=={cp.__version__}'])
         
         self._deps = list(deps)
@@ -286,11 +300,6 @@ class Isomap(BaseTransformer):
         if isinstance(dataset, DataFrame):
             session = dataset._session
             assert session is not None  # keep mypy happy
-            # Validate that key package version in user workspace are supported in snowflake conda channel
-            # If customer doesn't have package in conda channel, replace the ones have the closest versions
-            self._deps = pkg_version_utils.get_valid_pkg_versions_supported_in_snowflake_conda_channel(
-                pkg_versions=self._get_dependencies(), session=session, subproject=_SUBPROJECT)
-            
             # Specify input columns so column pruning will be enforced
             selected_cols = self._get_active_columns()
             if len(selected_cols) > 0:
@@ -318,7 +327,9 @@ class Isomap(BaseTransformer):
             label_cols=self.label_cols,
             sample_weight_col=self.sample_weight_col,
             autogenerated=self._autogenerated,
-            subproject=_SUBPROJECT
+            subproject=_SUBPROJECT,
+            use_external_memory_version=self._use_external_memory_version,
+            batch_size=self._batch_size,
         )
         self._sklearn_object = model_trainer.train()
         self._is_fitted = True
@@ -589,6 +600,22 @@ class Isomap(BaseTransformer):
                 # each row containing a list of values.
                 expected_dtype = "ARRAY"
 
+            # If we were unable to assign a type to this transform in the factory, infer the type here.
+            if expected_dtype == "":
+                # If this is a clustering transformer, if the number of output columns does not equal the number of clusters the response will be an "ARRAY"
+                if hasattr(self._sklearn_object, "n_clusters") and getattr(self._sklearn_object, "n_clusters") != len(self.output_cols):
+                    expected_dtype = "ARRAY"
+                # If this is a decomposition transformer, if the number of output columns does not equal the number of components the response will be an "ARRAY"
+                elif hasattr(self._sklearn_object, "n_components") and getattr(self._sklearn_object, "n_components") != len(self.output_cols):
+                    expected_dtype = "ARRAY"
+                else:
+                    output_types = [signature.as_snowpark_type() for signature in _infer_signature(dataset[self.input_cols], "output", use_snowflake_identifiers=True)]
+                    # We can only infer the output types from the input types if the following two statemetns are true:
+                    # 1) All of the output types are the same. Otherwise, we still have to fall back to variant because `_sklearn_inference` only accepts one type.
+                    # 2) The length of the input columns equals the length of the output columns. Otherwise the transform will likely result in an `ARRAY`.
+                    if all(x == output_types[0] for x in output_types) and len(output_types) == len(self.output_cols):
+                        expected_dtype = convert_sp_to_sf_type(output_types[0])
+            
             output_df = self._batch_inference(
                 dataset=dataset,
                 inference_method="transform",
@@ -604,8 +631,8 @@ class Isomap(BaseTransformer):
 
         return output_df
     
-    @available_if(original_estimator_has_callable("fit_predict"))  # type: ignore[misc]
-    def fit_predict(self, dataset: Union[DataFrame, pd.DataFrame]) -> npt.NDArray[Any]:
+    @available_if(_is_fit_predict_method_enabled())  # type: ignore[misc]
+    def fit_predict(self, dataset: Union[DataFrame, pd.DataFrame]) -> Union[Any, npt.NDArray[Any]]:
         """ Method not supported for this class.
 
 
@@ -618,13 +645,21 @@ class Isomap(BaseTransformer):
         Returns:
             Predicted dataset.
         """
-        if False:
-            self.fit(dataset)
-            assert self._sklearn_object is not None
-            labels : npt.NDArray[Any] = self._sklearn_object.labels_
-            return labels
-        else:
-            raise NotImplementedError
+        self.fit(dataset)
+        assert self._sklearn_object is not None
+        return self._sklearn_object.labels_
+
+
+    @available_if(_is_fit_transform_method_enabled())  # type: ignore[misc]
+    def fit_transform(self, dataset: Union[DataFrame, pd.DataFrame]) -> Union[Any, npt.NDArray[Any]]:
+        """ 
+        Returns:
+            Transformed dataset.
+        """
+        self.fit(dataset)
+        assert self._sklearn_object is not None
+        return self._sklearn_object.embedding_
+
 
     def _get_output_column_names(self, output_cols_prefix: str, output_cols: Optional[List[str]] = None) -> List[str]:
         """ Returns the list of output columns for predict_proba(), decision_function(), etc.. functions.

snowflake/ml/modeling/manifold/mds.py

@@ -54,6 +54,18 @@ _PROJECT = "ModelDevelopment"
 _SUBPROJECT = "".join([s.capitalize() for s in "sklearn.manifold".replace("sklearn.", "").split("_")])
 
 
+def _is_fit_predict_method_enabled() -> Callable[[Any], bool]:
+    def check(self: BaseTransformer) -> TypeGuard[Callable[..., object]]:
+        return False and callable(getattr(self._sklearn_object, "fit_predict", None))
+    return check
+
+
+def _is_fit_transform_method_enabled() -> Callable[[Any], bool]:
+    def check(self: BaseTransformer) -> TypeGuard[Callable[..., object]]:
+        return True and callable(getattr(self._sklearn_object, "fit_transform", None))
+    return check
+
+
 class MDS(BaseTransformer):
     r"""Multidimensional scaling
     For more details on this class, see [sklearn.manifold.MDS]
@@ -184,7 +196,9 @@ class MDS(BaseTransformer):
         self.set_label_cols(label_cols)
         self.set_passthrough_cols(passthrough_cols)
         self.set_drop_input_cols(drop_input_cols)
-        self.set_sample_weight_col(sample_weight_col)        
+        self.set_sample_weight_col(sample_weight_col)
+        self._use_external_memory_version = False
+        self._batch_size = -1        
         deps: Set[str] = set([f'numpy=={np.__version__}', f'scikit-learn=={sklearn.__version__}', f'cloudpickle=={cp.__version__}'])
         
         self._deps = list(deps)
@@ -269,11 +283,6 @@ class MDS(BaseTransformer):
         if isinstance(dataset, DataFrame):
             session = dataset._session
             assert session is not None  # keep mypy happy
-            # Validate that key package version in user workspace are supported in snowflake conda channel
-            # If customer doesn't have package in conda channel, replace the ones have the closest versions
-            self._deps = pkg_version_utils.get_valid_pkg_versions_supported_in_snowflake_conda_channel(
-                pkg_versions=self._get_dependencies(), session=session, subproject=_SUBPROJECT)
-            
             # Specify input columns so column pruning will be enforced
             selected_cols = self._get_active_columns()
             if len(selected_cols) > 0:
@@ -301,7 +310,9 @@ class MDS(BaseTransformer):
             label_cols=self.label_cols,
             sample_weight_col=self.sample_weight_col,
             autogenerated=self._autogenerated,
-            subproject=_SUBPROJECT
+            subproject=_SUBPROJECT,
+            use_external_memory_version=self._use_external_memory_version,
+            batch_size=self._batch_size,
         )
         self._sklearn_object = model_trainer.train()
         self._is_fitted = True
@@ -570,6 +581,22 @@ class MDS(BaseTransformer):
                 # each row containing a list of values.
                 expected_dtype = "ARRAY"
 
+            # If we were unable to assign a type to this transform in the factory, infer the type here.
+            if expected_dtype == "":
+                # If this is a clustering transformer, if the number of output columns does not equal the number of clusters the response will be an "ARRAY"
+                if hasattr(self._sklearn_object, "n_clusters") and getattr(self._sklearn_object, "n_clusters") != len(self.output_cols):
+                    expected_dtype = "ARRAY"
+                # If this is a decomposition transformer, if the number of output columns does not equal the number of components the response will be an "ARRAY"
+                elif hasattr(self._sklearn_object, "n_components") and getattr(self._sklearn_object, "n_components") != len(self.output_cols):
+                    expected_dtype = "ARRAY"
+                else:
+                    output_types = [signature.as_snowpark_type() for signature in _infer_signature(dataset[self.input_cols], "output", use_snowflake_identifiers=True)]
+                    # We can only infer the output types from the input types if the following two statemetns are true:
+                    # 1) All of the output types are the same. Otherwise, we still have to fall back to variant because `_sklearn_inference` only accepts one type.
+                    # 2) The length of the input columns equals the length of the output columns. Otherwise the transform will likely result in an `ARRAY`.
+                    if all(x == output_types[0] for x in output_types) and len(output_types) == len(self.output_cols):
+                        expected_dtype = convert_sp_to_sf_type(output_types[0])
+            
             output_df = self._batch_inference(
                 dataset=dataset,
                 inference_method="transform",
@@ -585,8 +612,8 @@ class MDS(BaseTransformer):
 
         return output_df
     
-    @available_if(original_estimator_has_callable("fit_predict"))  # type: ignore[misc]
-    def fit_predict(self, dataset: Union[DataFrame, pd.DataFrame]) -> npt.NDArray[Any]:
+    @available_if(_is_fit_predict_method_enabled())  # type: ignore[misc]
+    def fit_predict(self, dataset: Union[DataFrame, pd.DataFrame]) -> Union[Any, npt.NDArray[Any]]:
         """ Method not supported for this class.
 
 
@@ -599,13 +626,21 @@ class MDS(BaseTransformer):
         Returns:
             Predicted dataset.
         """
-        if False:
-            self.fit(dataset)
-            assert self._sklearn_object is not None
-            labels : npt.NDArray[Any] = self._sklearn_object.labels_
-            return labels
-        else:
-            raise NotImplementedError
+        self.fit(dataset)
+        assert self._sklearn_object is not None
+        return self._sklearn_object.labels_
+
+
+    @available_if(_is_fit_transform_method_enabled())  # type: ignore[misc]
+    def fit_transform(self, dataset: Union[DataFrame, pd.DataFrame]) -> Union[Any, npt.NDArray[Any]]:
+        """ 
+        Returns:
+            Transformed dataset.
+        """
+        self.fit(dataset)
+        assert self._sklearn_object is not None
+        return self._sklearn_object.embedding_
+
 
     def _get_output_column_names(self, output_cols_prefix: str, output_cols: Optional[List[str]] = None) -> List[str]:
         """ Returns the list of output columns for predict_proba(), decision_function(), etc.. functions.

snowflake/ml/modeling/manifold/spectral_embedding.py

@@ -54,6 +54,18 @@ _PROJECT = "ModelDevelopment"
 _SUBPROJECT = "".join([s.capitalize() for s in "sklearn.manifold".replace("sklearn.", "").split("_")])
 
 
+def _is_fit_predict_method_enabled() -> Callable[[Any], bool]:
+    def check(self: BaseTransformer) -> TypeGuard[Callable[..., object]]:
+        return False and callable(getattr(self._sklearn_object, "fit_predict", None))
+    return check
+
+
+def _is_fit_transform_method_enabled() -> Callable[[Any], bool]:
+    def check(self: BaseTransformer) -> TypeGuard[Callable[..., object]]:
+        return True and callable(getattr(self._sklearn_object, "fit_transform", None))
+    return check
+
+
 class SpectralEmbedding(BaseTransformer):
     r"""Spectral embedding for non-linear dimensionality reduction
     For more details on this class, see [sklearn.manifold.SpectralEmbedding]
@@ -188,7 +200,9 @@ class SpectralEmbedding(BaseTransformer):
         self.set_label_cols(label_cols)
         self.set_passthrough_cols(passthrough_cols)
         self.set_drop_input_cols(drop_input_cols)
-        self.set_sample_weight_col(sample_weight_col)        
+        self.set_sample_weight_col(sample_weight_col)
+        self._use_external_memory_version = False
+        self._batch_size = -1        
         deps: Set[str] = set([f'numpy=={np.__version__}', f'scikit-learn=={sklearn.__version__}', f'cloudpickle=={cp.__version__}'])
         
         self._deps = list(deps)
@@ -271,11 +285,6 @@ class SpectralEmbedding(BaseTransformer):
         if isinstance(dataset, DataFrame):
             session = dataset._session
             assert session is not None  # keep mypy happy
-            # Validate that key package version in user workspace are supported in snowflake conda channel
-            # If customer doesn't have package in conda channel, replace the ones have the closest versions
-            self._deps = pkg_version_utils.get_valid_pkg_versions_supported_in_snowflake_conda_channel(
-                pkg_versions=self._get_dependencies(), session=session, subproject=_SUBPROJECT)
-            
             # Specify input columns so column pruning will be enforced
             selected_cols = self._get_active_columns()
             if len(selected_cols) > 0:
@@ -303,7 +312,9 @@ class SpectralEmbedding(BaseTransformer):
             label_cols=self.label_cols,
             sample_weight_col=self.sample_weight_col,
             autogenerated=self._autogenerated,
-            subproject=_SUBPROJECT
+            subproject=_SUBPROJECT,
+            use_external_memory_version=self._use_external_memory_version,
+            batch_size=self._batch_size,
         )
         self._sklearn_object = model_trainer.train()
         self._is_fitted = True
@@ -572,6 +583,22 @@ class SpectralEmbedding(BaseTransformer):
                 # each row containing a list of values.
                 expected_dtype = "ARRAY"
 
+            # If we were unable to assign a type to this transform in the factory, infer the type here.
+            if expected_dtype == "":
+                # If this is a clustering transformer, if the number of output columns does not equal the number of clusters the response will be an "ARRAY"
+                if hasattr(self._sklearn_object, "n_clusters") and getattr(self._sklearn_object, "n_clusters") != len(self.output_cols):
+                    expected_dtype = "ARRAY"
+                # If this is a decomposition transformer, if the number of output columns does not equal the number of components the response will be an "ARRAY"
+                elif hasattr(self._sklearn_object, "n_components") and getattr(self._sklearn_object, "n_components") != len(self.output_cols):
+                    expected_dtype = "ARRAY"
+                else:
+                    output_types = [signature.as_snowpark_type() for signature in _infer_signature(dataset[self.input_cols], "output", use_snowflake_identifiers=True)]
+                    # We can only infer the output types from the input types if the following two statemetns are true:
+                    # 1) All of the output types are the same. Otherwise, we still have to fall back to variant because `_sklearn_inference` only accepts one type.
+                    # 2) The length of the input columns equals the length of the output columns. Otherwise the transform will likely result in an `ARRAY`.
+                    if all(x == output_types[0] for x in output_types) and len(output_types) == len(self.output_cols):
+                        expected_dtype = convert_sp_to_sf_type(output_types[0])
+            
             output_df = self._batch_inference(
                 dataset=dataset,
                 inference_method="transform",
@@ -587,8 +614,8 @@ class SpectralEmbedding(BaseTransformer):
 
         return output_df
     
-    @available_if(original_estimator_has_callable("fit_predict"))  # type: ignore[misc]
-    def fit_predict(self, dataset: Union[DataFrame, pd.DataFrame]) -> npt.NDArray[Any]:
+    @available_if(_is_fit_predict_method_enabled())  # type: ignore[misc]
+    def fit_predict(self, dataset: Union[DataFrame, pd.DataFrame]) -> Union[Any, npt.NDArray[Any]]:
         """ Method not supported for this class.
 
 
@@ -601,13 +628,21 @@ class SpectralEmbedding(BaseTransformer):
         Returns:
             Predicted dataset.
         """
-        if False:
-            self.fit(dataset)
-            assert self._sklearn_object is not None
-            labels : npt.NDArray[Any] = self._sklearn_object.labels_
-            return labels
-        else:
-            raise NotImplementedError
+        self.fit(dataset)
+        assert self._sklearn_object is not None
+        return self._sklearn_object.labels_
+
+
+    @available_if(_is_fit_transform_method_enabled())  # type: ignore[misc]
+    def fit_transform(self, dataset: Union[DataFrame, pd.DataFrame]) -> Union[Any, npt.NDArray[Any]]:
+        """ 
+        Returns:
+            Transformed dataset.
+        """
+        self.fit(dataset)
+        assert self._sklearn_object is not None
+        return self._sklearn_object.embedding_
+
 
     def _get_output_column_names(self, output_cols_prefix: str, output_cols: Optional[List[str]] = None) -> List[str]:
         """ Returns the list of output columns for predict_proba(), decision_function(), etc.. functions.

snowflake/ml/modeling/manifold/tsne.py

@@ -54,6 +54,18 @@ _PROJECT = "ModelDevelopment"
 _SUBPROJECT = "".join([s.capitalize() for s in "sklearn.manifold".replace("sklearn.", "").split("_")])
 
 
+def _is_fit_predict_method_enabled() -> Callable[[Any], bool]:
+    def check(self: BaseTransformer) -> TypeGuard[Callable[..., object]]:
+        return False and callable(getattr(self._sklearn_object, "fit_predict", None))
+    return check
+
+
+def _is_fit_transform_method_enabled() -> Callable[[Any], bool]:
+    def check(self: BaseTransformer) -> TypeGuard[Callable[..., object]]:
+        return True and callable(getattr(self._sklearn_object, "fit_transform", None))
+    return check
+
+
 class TSNE(BaseTransformer):
     r"""T-distributed Stochastic Neighbor Embedding
     For more details on this class, see [sklearn.manifold.TSNE]
@@ -240,7 +252,9 @@ class TSNE(BaseTransformer):
         self.set_label_cols(label_cols)
         self.set_passthrough_cols(passthrough_cols)
         self.set_drop_input_cols(drop_input_cols)
-        self.set_sample_weight_col(sample_weight_col)        
+        self.set_sample_weight_col(sample_weight_col)
+        self._use_external_memory_version = False
+        self._batch_size = -1        
         deps: Set[str] = set([f'numpy=={np.__version__}', f'scikit-learn=={sklearn.__version__}', f'cloudpickle=={cp.__version__}'])
         
         self._deps = list(deps)
@@ -330,11 +344,6 @@ class TSNE(BaseTransformer):
         if isinstance(dataset, DataFrame):
             session = dataset._session
             assert session is not None  # keep mypy happy
-            # Validate that key package version in user workspace are supported in snowflake conda channel
-            # If customer doesn't have package in conda channel, replace the ones have the closest versions
-            self._deps = pkg_version_utils.get_valid_pkg_versions_supported_in_snowflake_conda_channel(
-                pkg_versions=self._get_dependencies(), session=session, subproject=_SUBPROJECT)
-            
             # Specify input columns so column pruning will be enforced
             selected_cols = self._get_active_columns()
             if len(selected_cols) > 0:
@@ -362,7 +371,9 @@ class TSNE(BaseTransformer):
             label_cols=self.label_cols,
             sample_weight_col=self.sample_weight_col,
             autogenerated=self._autogenerated,
-            subproject=_SUBPROJECT
+            subproject=_SUBPROJECT,
+            use_external_memory_version=self._use_external_memory_version,
+            batch_size=self._batch_size,
         )
         self._sklearn_object = model_trainer.train()
         self._is_fitted = True
@@ -631,6 +642,22 @@ class TSNE(BaseTransformer):
                 # each row containing a list of values.
                 expected_dtype = "ARRAY"
 
+            # If we were unable to assign a type to this transform in the factory, infer the type here.
+            if expected_dtype == "":
+                # If this is a clustering transformer, if the number of output columns does not equal the number of clusters the response will be an "ARRAY"
+                if hasattr(self._sklearn_object, "n_clusters") and getattr(self._sklearn_object, "n_clusters") != len(self.output_cols):
+                    expected_dtype = "ARRAY"
+                # If this is a decomposition transformer, if the number of output columns does not equal the number of components the response will be an "ARRAY"
+                elif hasattr(self._sklearn_object, "n_components") and getattr(self._sklearn_object, "n_components") != len(self.output_cols):
+                    expected_dtype = "ARRAY"
+                else:
+                    output_types = [signature.as_snowpark_type() for signature in _infer_signature(dataset[self.input_cols], "output", use_snowflake_identifiers=True)]
+                    # We can only infer the output types from the input types if the following two statemetns are true:
+                    # 1) All of the output types are the same. Otherwise, we still have to fall back to variant because `_sklearn_inference` only accepts one type.
+                    # 2) The length of the input columns equals the length of the output columns. Otherwise the transform will likely result in an `ARRAY`.
+                    if all(x == output_types[0] for x in output_types) and len(output_types) == len(self.output_cols):
+                        expected_dtype = convert_sp_to_sf_type(output_types[0])
+            
             output_df = self._batch_inference(
                 dataset=dataset,
                 inference_method="transform",
@@ -646,8 +673,8 @@ class TSNE(BaseTransformer):
 
         return output_df
     
-    @available_if(original_estimator_has_callable("fit_predict"))  # type: ignore[misc]
-    def fit_predict(self, dataset: Union[DataFrame, pd.DataFrame]) -> npt.NDArray[Any]:
+    @available_if(_is_fit_predict_method_enabled())  # type: ignore[misc]
+    def fit_predict(self, dataset: Union[DataFrame, pd.DataFrame]) -> Union[Any, npt.NDArray[Any]]:
         """ Method not supported for this class.
 
 
@@ -660,13 +687,21 @@ class TSNE(BaseTransformer):
         Returns:
             Predicted dataset.
         """
-        if False:
-            self.fit(dataset)
-            assert self._sklearn_object is not None
-            labels : npt.NDArray[Any] = self._sklearn_object.labels_
-            return labels
-        else:
-            raise NotImplementedError
+        self.fit(dataset)
+        assert self._sklearn_object is not None
+        return self._sklearn_object.labels_
+
+
+    @available_if(_is_fit_transform_method_enabled())  # type: ignore[misc]
+    def fit_transform(self, dataset: Union[DataFrame, pd.DataFrame]) -> Union[Any, npt.NDArray[Any]]:
+        """ 
+        Returns:
+            Transformed dataset.
+        """
+        self.fit(dataset)
+        assert self._sklearn_object is not None
+        return self._sklearn_object.embedding_
+
 
     def _get_output_column_names(self, output_cols_prefix: str, output_cols: Optional[List[str]] = None) -> List[str]:
         """ Returns the list of output columns for predict_proba(), decision_function(), etc.. functions.

snowflake/ml/modeling/metrics/classification.py

@@ -228,16 +228,15 @@ def _register_confusion_matrix_computer(*, session: snowpark.Session, statement_
     Returns:
         Name of the UDTF.
     """
+    batch_size = metrics_utils.BATCH_SIZE
 
     class ConfusionMatrixComputer:
-        BATCH_SIZE = 1000
-
         def __init__(self) -> None:
             self._initialized = False
             self._confusion_matrix = np.zeros((1, 1))
-            # 2d array containing a batch of input rows. A batch contains self.BATCH_SIZE rows.
+            # 2d array containing a batch of input rows. A batch contains metrics_utils.BATCH_SIZE rows.
             # [sample_weight, y_true, y_pred]
-            self._batched_rows = np.zeros((self.BATCH_SIZE, 1))
+            self._batched_rows = np.zeros((batch_size, 1))
             # Number of columns in the dataset.
             self._n_cols = -1
             # Running count of number of rows added to self._batched_rows.
@@ -255,7 +254,7 @@ def _register_confusion_matrix_computer(*, session: snowpark.Session, statement_
             # 1. Initialize variables.
             if not self._initialized:
                 self._n_cols = len(input_row)
-                self._batched_rows = np.zeros((self.BATCH_SIZE, self._n_cols))
+                self._batched_rows = np.zeros((batch_size, self._n_cols))
                 self._n_label = n_label
                 self._confusion_matrix = np.zeros((self._n_label, self._n_label))
             self._initialized = True
@@ -264,7 +263,7 @@ def _register_confusion_matrix_computer(*, session: snowpark.Session, statement_
             self._cur_count += 1
 
             # 2. Compute incremental confusion matrix for the batch.
-            if self._cur_count >= self.BATCH_SIZE:
+            if self._cur_count >= batch_size:
                 self.update_confusion_matrix()
                 self._cur_count = 0
 

snowflake/ml/modeling/metrics/metrics_utils.py

@@ -15,6 +15,7 @@ from snowflake.snowpark import Session, functions as F, types as T
 
 LABEL = "LABEL"
 INDEX = "INDEX"
+BATCH_SIZE = 1000
 
 
 def register_accumulator_udtf(*, session: Session, statement_params: Dict[str, Any]) -> str:
@@ -82,7 +83,7 @@ def register_sharded_dot_sum_computer(*, session: Session, statement_params: Dic
         """This class is registered as a UDTF and computes the sum and dot product
         of columns for each partition of rows. The computations across all the partitions happens
         in parallel using the nodes in the warehouse. In order to avoid keeping the entire partition
-        in memory, we batch the rows (size is 1000) and maintain a running sum and dot prod in self._sum_by_count,
+        in memory, we batch the rows and maintain a running sum and dot prod in self._sum_by_count,
         self._sum_by_countd and self._dot_prod respectively. We return these at the end of the partition.
         """
 
@@ -95,7 +96,7 @@ def register_sharded_dot_sum_computer(*, session: Session, statement_params: Dic
             # delta degree of freedom
             self._ddof = 0
             # Setting the batch size to 1000 based on experimentation. Can be fine tuned later.
-            self._batch_size = 1000
+            self._batch_size = BATCH_SIZE
             # 2d array containing a batch of input rows. A batch contains self._batch_size rows.
             self._batched_rows = np.zeros((self._batch_size, 1))
             # 1d array of length = # of cols. Contains sum(col/count) for each column.
@@ -224,7 +225,7 @@ def check_label_columns(
         TypeError: `y_true_col_names` and `y_pred_col_names` are of different types.
         ValueError: Multilabel `y_true_col_names` and `y_pred_col_names` are of different lengths.
     """
-    if type(y_true_col_names) != type(y_pred_col_names):
+    if type(y_true_col_names) is not type(y_pred_col_names):
         raise TypeError(
             "Label columns should be of the same type."
             f"Got y_true_col_names={type(y_true_col_names)} vs y_pred_col_names={type(y_pred_col_names)}."
@@ -300,6 +301,7 @@ def validate_average_pos_label(average: Optional[str] = None, pos_label: Union[s
             "average != 'binary' (got %r). You may use "
             "labels=[pos_label] to specify a single positive class." % (pos_label, average),
             UserWarning,
+            stacklevel=2,
         )