snowflake-ml-python 1.4.0__py3-none-any.whl → 1.5.0__py3-none-any.whl

snowflake/ml/modeling/preprocessing/binarizer.py

@@ -86,11 +86,7 @@ class Binarizer(base.BaseTransformer):
         """
         super()._reset()
 
-    @telemetry.send_api_usage_telemetry(
-        project=base.PROJECT,
-        subproject=base.SUBPROJECT,
-    )
-    def fit(self, dataset: Union[snowpark.DataFrame, pd.DataFrame]) -> "Binarizer":
+    def _fit(self, dataset: Union[snowpark.DataFrame, pd.DataFrame]) -> "Binarizer":
         """
         This is a stateless transformer, so there is nothing to fit. Validates the transformer arguments.
         Returns the transformer instance.

snowflake/ml/modeling/preprocessing/k_bins_discretizer.py

@@ -146,11 +146,7 @@ class KBinsDiscretizer(base.BaseTransformer):
         self.bin_edges_: Optional[npt.NDArray[np.float32]] = None
         self.n_bins_: Optional[npt.NDArray[np.int32]] = None
 
-    @telemetry.send_api_usage_telemetry(
-        project=base.PROJECT,
-        subproject=base.SUBPROJECT,
-    )
-    def fit(self, dataset: Union[snowpark.DataFrame, pd.DataFrame]) -> KBinsDiscretizer:
+    def _fit(self, dataset: Union[snowpark.DataFrame, pd.DataFrame]) -> KBinsDiscretizer:
         """
         Fit KBinsDiscretizer with dataset.
 

snowflake/ml/modeling/preprocessing/label_encoder.py

@@ -87,11 +87,7 @@ class LabelEncoder(base.BaseTransformer):
             self._ordinal_encoder = None
             self.classes_ = None
 
-    @telemetry.send_api_usage_telemetry(
-        project=base.PROJECT,
-        subproject=base.SUBPROJECT,
-    )
-    def fit(self, dataset: Union[snowpark.DataFrame, pd.DataFrame]) -> "LabelEncoder":
+    def _fit(self, dataset: Union[snowpark.DataFrame, pd.DataFrame]) -> "LabelEncoder":
         """
         Fit label encoder with label column in dataset.
 

snowflake/ml/modeling/preprocessing/max_abs_scaler.py

@@ -106,11 +106,7 @@ class MaxAbsScaler(base.BaseTransformer):
         self.scale_ = {}
         self.max_abs_ = {}
 
-    @telemetry.send_api_usage_telemetry(
-        project=base.PROJECT,
-        subproject=base.SUBPROJECT,
-    )
-    def fit(self, dataset: Union[snowpark.DataFrame, pd.DataFrame]) -> "MaxAbsScaler":
+    def _fit(self, dataset: Union[snowpark.DataFrame, pd.DataFrame]) -> "MaxAbsScaler":
         """
         Compute the maximum absolute value to be used for later scaling.
 

snowflake/ml/modeling/preprocessing/min_max_scaler.py

@@ -138,11 +138,7 @@ class MinMaxScaler(base.BaseTransformer):
                         ),
                     )
 
-    @telemetry.send_api_usage_telemetry(
-        project=base.PROJECT,
-        subproject=base.SUBPROJECT,
-    )
-    def fit(self, dataset: Union[snowpark.DataFrame, pd.DataFrame]) -> "MinMaxScaler":
+    def _fit(self, dataset: Union[snowpark.DataFrame, pd.DataFrame]) -> "MinMaxScaler":
         """
         Compute min and max values of the dataset.
 
@@ -175,11 +171,13 @@ class MinMaxScaler(base.BaseTransformer):
         sklearn_scaler.fit(dataset[self.input_cols])
 
         for i, input_col in enumerate(self.input_cols):
-            self.min_[input_col] = float(sklearn_scaler.min_[i])
-            self.scale_[input_col] = float(sklearn_scaler.scale_[i])
-            self.data_min_[input_col] = float(sklearn_scaler.data_min_[i])
-            self.data_max_[input_col] = float(sklearn_scaler.data_max_[i])
-            self.data_range_[input_col] = float(sklearn_scaler.data_range_[i])
+            self.min_[input_col] = _utils.to_float_if_valid(sklearn_scaler.min_[i], input_col, "min_")
+            self.scale_[input_col] = _utils.to_float_if_valid(sklearn_scaler.scale_[i], input_col, "scale_")
+            self.data_min_[input_col] = _utils.to_float_if_valid(sklearn_scaler.data_min_[i], input_col, "data_min_")
+            self.data_max_[input_col] = _utils.to_float_if_valid(sklearn_scaler.data_max_[i], input_col, "data_max_")
+            self.data_range_[input_col] = _utils.to_float_if_valid(
+                sklearn_scaler.data_range_[i], input_col, "data_range_"
+            )
 
     def _fit_snowpark(self, dataset: snowpark.DataFrame) -> None:
         self._check_input_column_types(dataset)
@@ -189,8 +187,8 @@ class MinMaxScaler(base.BaseTransformer):
         for input_col in self.input_cols:
             numeric_stats = computed_states[input_col]
 
-            data_min = float(numeric_stats[_utils.NumericStatistics.MIN])
-            data_max = float(numeric_stats[_utils.NumericStatistics.MAX])
+            data_min = _utils.to_float_if_valid(numeric_stats[_utils.NumericStatistics.MIN], input_col, "data_min_")
+            data_max = _utils.to_float_if_valid(numeric_stats[_utils.NumericStatistics.MAX], input_col, "data_max_")
             data_range = data_max - data_min
             self.scale_[input_col] = (
                 self.feature_range[1] - self.feature_range[0]

snowflake/ml/modeling/preprocessing/normalizer.py

@@ -70,11 +70,7 @@ class Normalizer(base.BaseTransformer):
         """
         pass
 
-    @telemetry.send_api_usage_telemetry(
-        project=base.PROJECT,
-        subproject=base.SUBPROJECT,
-    )
-    def fit(self, dataset: Union[snowpark.DataFrame, pd.DataFrame]) -> "Normalizer":
+    def _fit(self, dataset: Union[snowpark.DataFrame, pd.DataFrame]) -> "Normalizer":
         """
         Does nothing, because the normalizer is a stateless transformer.
 

snowflake/ml/modeling/preprocessing/one_hot_encoder.py

@@ -267,11 +267,7 @@ class OneHotEncoder(base.BaseTransformer):
         if hasattr(self, "_state_pandas"):
             del self._state_pandas
 
-    @telemetry.send_api_usage_telemetry(
-        project=base.PROJECT,
-        subproject=base.SUBPROJECT,
-    )
-    def fit(self, dataset: Union[snowpark.DataFrame, pd.DataFrame]) -> "OneHotEncoder":
+    def _fit(self, dataset: Union[snowpark.DataFrame, pd.DataFrame]) -> "OneHotEncoder":
         """
         Fit OneHotEncoder to dataset.
 
@@ -836,6 +832,18 @@ class OneHotEncoder(base.BaseTransformer):
 
         # columns: COLUMN_NAME, CATEGORY, COUNT, FITTED_CATEGORY, ENCODING, N_FEATURES_OUT, ENCODED_VALUE, OUTPUT_CATs
         assert dataset._session is not None
+
+        def convert_to_string_excluding_nan(item: Any) -> Union[None, str]:
+            if pd.isna(item):
+                return None  # or np.nan if you prefer to keep as NaN
+            else:
+                return str(item)
+
+        # In case of fitting with pandas dataframe and transforming with snowpark dataframe
+        # state_pandas cannot recognize the datatype of _CATEGORY and _FITTED_CATEGORY column
+        # Therefore, apply the convert_to_string_excluding_nan function to _CATEGORY and _FITTED_CATEGORY
+        state_pandas[[_CATEGORY]] = state_pandas[[_CATEGORY]].applymap(convert_to_string_excluding_nan)
+        state_pandas[[_FITTED_CATEGORY]] = state_pandas[[_FITTED_CATEGORY]].applymap(convert_to_string_excluding_nan)
         state_df = dataset._session.create_dataframe(state_pandas)
 
         transformed_dataset = dataset

snowflake/ml/modeling/preprocessing/ordinal_encoder.py

@@ -171,11 +171,7 @@ class OrdinalEncoder(base.BaseTransformer):
         if hasattr(self, "_state_pandas"):
             del self._state_pandas
 
-    @telemetry.send_api_usage_telemetry(
-        project=base.PROJECT,
-        subproject=base.SUBPROJECT,
-    )
-    def fit(self, dataset: Union[snowpark.DataFrame, pd.DataFrame]) -> "OrdinalEncoder":
+    def _fit(self, dataset: Union[snowpark.DataFrame, pd.DataFrame]) -> "OrdinalEncoder":
         """
         Fit the OrdinalEncoder to dataset.