#286 data.apply(lambda x: x) -> data.obj.copy()

src/pytimetk/core/summarize_by_time.py

@@ -328,10 +328,10 @@ def _summarize_by_time_polars(
         grouped = True
         # Extract names from groupby object
         groups = data.grouper.names  # This can be a list of group names
-
+        
         # Convert the GroupBy object into a Polars DataFrame
         df_pl = (
-            pl.from_pandas(data.apply(lambda x: x))
+            pl.from_pandas(data.obj.copy())
                  .group_by(groups, maintain_order=True)
                  .agg(pl.all().sort_by(date_column))
         )

src/pytimetk/feature_engineering/diffs.py

@@ -196,7 +196,7 @@ def _augment_diffs_polars(
 
     if isinstance(data, pd.core.groupby.generic.DataFrameGroupBy):
         # Data is a GroupBy object, use apply to get a DataFrame
-        pandas_df = data.apply(lambda x: x)
+        pandas_df = data.obj.copy()
     elif isinstance(data, pd.DataFrame):
         # Data is already a DataFrame
         pandas_df = data

src/pytimetk/feature_engineering/expanding.py

@@ -418,20 +418,17 @@ def _augment_expanding_polars(
     Augments the given dataframe with expanding calculations using the Polars library.
     """
 
-    # Create a fresh copy of the data, leaving the original untouched
-    data_copy = data.copy() if isinstance(data, pd.DataFrame) else data.obj.copy()
-
     # Retrieve the group column names if the input data is a GroupBy object
     if isinstance(data, pd.core.groupby.generic.DataFrameGroupBy):
         group_names = data.grouper.names
     else: 
         group_names = None
 
     # Convert data into a Pandas DataFrame format for processing
-    if isinstance(data_copy, pd.core.groupby.generic.DataFrameGroupBy):
-        pandas_df = data_copy.apply(lambda x: x)
-    elif isinstance(data_copy, pd.DataFrame):
-        pandas_df = data_copy
+    if isinstance(data, pd.core.groupby.generic.DataFrameGroupBy):
+        pandas_df = data.obj.copy()
+    elif isinstance(data, pd.DataFrame):
+        pandas_df = data.copy()
     else:
         raise ValueError("Data must be a Pandas DataFrame or Pandas GroupBy object.")
 

src/pytimetk/feature_engineering/fourier.py

@@ -120,7 +120,7 @@ def augment_fourier(
         raise TypeError(f"Invalid periods specification: type: {type(periods)}. Please use int, tuple, or list.")
 
     if isinstance(data, pd.core.groupby.generic.DataFrameGroupBy):  
-        data = data.apply(lambda x: x).reset_index(drop=True)  
+        data = data.obj.copy().reset_index(drop=True)  
 
     # Reduce memory usage
     if reduce_memory:
@@ -156,7 +156,6 @@ def _augment_fourier_pandas(
 ) -> pd.DataFrame:
 
     df = data.copy()
-    # df.sort_values(by=[date_column], inplace=True)
 
     scale_factor = date_to_seq_scale_factor(df, date_column).iloc[0].total_seconds()
     if scale_factor == 0:

src/pytimetk/feature_engineering/hilbert.py

@@ -280,7 +280,7 @@ def apply_hilbert(pl_df):
 
         # Convert the GroupBy object into a Polars DataFrame
         df_pl = (
-            pl.from_pandas(data.apply(lambda x: x))
+            pl.from_pandas(data.obj.copy())
                  .group_by(groups, maintain_order=True)
                  .agg(pl.all().sort_by(date_column))
         )

src/pytimetk/feature_engineering/holiday_signature.py

@@ -228,7 +228,7 @@ def _augment_holiday_signature_pandas(
 
     if isinstance(data, pd.core.groupby.generic.DataFrameGroupBy):
         # Data is a GroupBy object, use apply to get a DataFrame
-        pandas_df = data.apply(lambda x: x)
+        pandas_df = data.obj.copy()
     elif isinstance(data, pd.DataFrame):
         # Data is already a DataFrame
         pandas_df = data
@@ -306,7 +306,7 @@ def _augment_holiday_signature_polars(
 
     if isinstance(data, pd.core.groupby.generic.DataFrameGroupBy):
         # Data is a GroupBy object, use apply to get a DataFrame
-        pandas_df = data.apply(lambda x: x)
+        pandas_df = data.obj.copy()
     elif isinstance(data, pd.DataFrame):
         # Data is already a DataFrame
         pandas_df = data

src/pytimetk/feature_engineering/lags.py

@@ -162,8 +162,6 @@ def _augment_lags_pandas(
 
         df = data.copy()
 
-        df.sort_values(by=[date_column], inplace=True)
-
         for col in value_column:
             for lag in lags:
                 df[f'{col}_lag_{lag}'] = df[col].shift(lag)
@@ -177,8 +175,6 @@ def _augment_lags_pandas(
 
         df = data.copy()
 
-        df.sort_values(by=[*group_names, date_column], inplace=True)
-
         for col in value_column:
             for lag in lags:
                 df[f'{col}_lag_{lag}'] = df.groupby(group_names)[col].shift(lag)
@@ -194,7 +190,7 @@ def _augment_lags_polars(
 
     if isinstance(data, pd.core.groupby.generic.DataFrameGroupBy):
         # Data is a GroupBy object, use apply to get a DataFrame
-        pandas_df = data.apply(lambda x: x)
+        pandas_df = data.obj
     elif isinstance(data, pd.DataFrame):
         # Data is already a DataFrame
         pandas_df = data

src/pytimetk/feature_engineering/leads.py

@@ -193,7 +193,7 @@ def _augment_leads_polars(
 ) -> pl.DataFrame:
     if isinstance(data, pd.core.groupby.generic.DataFrameGroupBy):
         # Data is a GroupBy object, use apply to get a DataFrame
-        pandas_df = data.apply(lambda x: x)
+        pandas_df = data.obj.copy()
     elif isinstance(data, pd.DataFrame):
         # Data is already a DataFrame
         pandas_df = data

src/pytimetk/feature_engineering/rolling.py

@@ -277,7 +277,7 @@ def _augment_rolling_pandas(
     # Group data if it's a GroupBy object; otherwise, prepare it for the rolling calculations
     if isinstance(data, pd.core.groupby.generic.DataFrameGroupBy):
         group_names = data.grouper.names
-        grouped = data_copy.sort_values(by=[*group_names, date_column]).groupby(group_names)
+        grouped = data_copy.groupby(group_names)
 
         # Check if the data is grouped and threads are set to 1. If true, handle it without parallel processing.
         if threads == 1:
@@ -408,8 +408,6 @@ def _augment_rolling_polars(
     **kwargs,
 ) -> pd.DataFrame:
 
-    # Create a fresh copy of the data, leaving the original untouched
-    data_copy = data.copy() if isinstance(data, pd.DataFrame) else data.obj.copy()
 
     # Retrieve the group column names if the input data is a GroupBy object
     if isinstance(data, pd.core.groupby.generic.DataFrameGroupBy):
@@ -418,10 +416,10 @@ def _augment_rolling_polars(
         group_names = None
 
     # Convert data into a Pandas DataFrame format for processing
-    if isinstance(data_copy, pd.core.groupby.generic.DataFrameGroupBy):
-        pandas_df = data_copy.apply(lambda x: x)
-    elif isinstance(data_copy, pd.DataFrame):
-        pandas_df = data_copy
+    if isinstance(data, pd.core.groupby.generic.DataFrameGroupBy):
+        pandas_df = data.obj.copy()
+    elif isinstance(data, pd.DataFrame):
+        pandas_df = data.copy()
     else:
         raise ValueError("Data must be a Pandas DataFrame or Pandas GroupBy object.")
 

src/pytimetk/finance/cmo.py

@@ -303,9 +303,6 @@ def _calculate_cmo_polars(series: pl.Series, period=14):
     delta = series.diff()
 
     # Separate gains and losses    
-    # gains = delta.apply(lambda x: x if x > 0 else 0)
-    # losses = delta.apply(lambda x: -x if x < 0 else 0)
-
     gains = pl.when(delta > 0).then(delta).otherwise(0)
     losses = pl.when(delta <= 0).then(-delta).otherwise(0)
 

src/pytimetk/finance/rsi.py

@@ -225,7 +225,7 @@ def _augment_rsi_polars(
 
     if isinstance(data, pd.core.groupby.generic.DataFrameGroupBy):
         # Data is a GroupBy object, use apply to get a DataFrame
-        pandas_df = data.apply(lambda x: x)
+        pandas_df = data.obj.copy()
     elif isinstance(data, pd.DataFrame):
         # Data is already a DataFrame
         pandas_df = data.copy()

tests/test_get_frequency_summary.py

@@ -15,6 +15,7 @@
     ('Y', False, 'A-DEC' , '365 days', 1.0, 'Y'),
     ('Y', True,  'Y'     , '365 days', 1.0, 'Y')
 ])
+
 def test_correct_frequency_inference(
     freq, regular, inferred_unit, median_timedelta, median_scale, median_unit
 ):