AnomalyLab 0.2.8__tar.gz → 0.3.0__tar.gz

{anomalylab-0.2.8 → anomalylab-0.3.0/AnomalyLab.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: AnomalyLab
-Version: 0.2.8
+Version: 0.3.0
 Summary: A Python package for empirical asset pricing analysis.
 Author: FinPhd
 Classifier: Programming Language :: Python :: 3

{anomalylab-0.2.8/AnomalyLab.egg-info → anomalylab-0.3.0}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: AnomalyLab
-Version: 0.2.8
+Version: 0.3.0
 Summary: A Python package for empirical asset pricing analysis.
 Author: FinPhd
 Classifier: Programming Language :: Python :: 3

{anomalylab-0.2.8 → anomalylab-0.3.0}/anomalylab/core/core.py

@@ -19,7 +19,7 @@ from anomalylab.visualization import FormatExcel
 
 @dataclass
 class Panel:
-    _df: pd.DataFrame = field(repr=False)
+    _df: DataFrame = field(repr=False)
     name: Optional[str] = None
     id: str = "permno"
     time: str = "date"
@@ -27,6 +27,7 @@ class Panel:
     ret: str = "return"
     classifications: Optional[list[str] | str] = None
     drop_all_chars_missing: bool = False
+    is_copy: bool = False
 
     def __post_init__(self) -> None:
         self.panel_data: PanelData = PanelData(
@@ -38,6 +39,7 @@ class Panel:
             ret=self.ret,
             classifications=self.classifications,
             drop_all_chars_missing=self.drop_all_chars_missing,
+            is_copy=self.is_copy,
         )
         self._normalize_processor = None
         self._fillna_processor = None
@@ -260,7 +262,7 @@ class Panel:
         draw: bool = False,
         path: Optional[str] = None,
         decimal: Optional[int] = None,
-    ) -> pd.DataFrame:
+    ) -> DataFrame:
         return self.persistence_processor.transition_matrix(
             var=var,
             group=group,
@@ -304,7 +306,7 @@ class Panel:
         factors_series: Optional[TimeSeries] = None,
         pivot: bool = True,
         format: bool = False,
-        type: str = "dependent",
+        sort_type: str = "dependent",
         decimal: Optional[int] = None,
         factor_return: bool = False,
     ) -> tuple:
@@ -317,7 +319,7 @@ class Panel:
             core_g=core_g,
             pivot=pivot,
             format=format,
-            type=type,
+            sort_type=sort_type,
             decimal=decimal,
             factor_return=factor_return,
         )
@@ -372,7 +374,11 @@ if __name__ == "__main__":
     }
 
     panel = Panel(
-        df, name="Stocks", classifications="industry", drop_all_chars_missing=True
+        df,
+        name="Stocks",
+        classifications="industry",
+        drop_all_chars_missing=True,
+        is_copy=False,
     )
     time_series: TimeSeries = TimeSeries(df=ts, name="Factor Series")
     pp(panel)

{anomalylab-0.2.8 → anomalylab-0.3.0}/anomalylab/empirical/correlation.py

@@ -74,7 +74,7 @@ class Correlation(Empirical):
             )
             is_upper = False  # Switch to lower triangle for the next method
 
-        return pd.DataFrame(data=merged_corr, index=columns, columns=columns).map(
+        return DataFrame(data=merged_corr, index=columns, columns=columns).map(
             func=round_to_string,
             decimal=decimal or self.decimal,  # Round results to specified decimals
         )

{anomalylab-0.2.8 → anomalylab-0.3.0}/anomalylab/empirical/fm_regression.py

@@ -131,11 +131,11 @@ class FamaMacBethRegression(Empirical):
                 coefs = results.params
                 coefs[self.time] = time
                 coef_df.append(coefs)
-            coef_df = pd.DataFrame(coef_df)
+            coef_df = DataFrame(coef_df)
             coef_df = coef_df[
                 [self.time] + [col for col in coef_df.columns if col != self.time]
             ]
-            return pd.DataFrame(coef_df)
+            return DataFrame(coef_df)
 
         # Fama-MacBeth regression with Newey-West adjustment
         fmb = FamaMacBeth(
@@ -207,7 +207,7 @@ class FamaMacBethRegression(Empirical):
         Returns:
             Series: Formatted regression results including parameters, t-values, and statistics.
         """
-        result: Series = pd.DataFrame(
+        result: Series = DataFrame(
             data={
                 "params": reg_result["params"].map(
                     arg=lambda x: round_to_string(value=x, decimal=decimal)

{anomalylab-0.2.8 → anomalylab-0.3.0}/anomalylab/empirical/persistence.py

@@ -29,7 +29,7 @@ class Persistence(Empirical):
         no_process_columns: Columns = None,
         process_all_characteristics: bool = True,
         decimal: Optional[int] = None,
-    ) -> pd.DataFrame:
+    ) -> DataFrame:
         """
         Computes average persistence (autocorrelation) for specified columns over defined time periods.
 
@@ -48,7 +48,7 @@ class Persistence(Empirical):
                 Defaults to None.
 
         Returns:
-            pd.DataFrame: A DataFrame containing the average persistence for specified columns.
+            DataFrame: A DataFrame containing the average persistence for specified columns.
 
         Note:
             The resulting DataFrame contains the average correlations for each lag, formatted to the
@@ -90,7 +90,7 @@ class Persistence(Empirical):
                 all_monthly_corrs.extend(monthly_corrs)
 
             # Convert to DataFrame
-            all_monthly_corrs_df = pd.DataFrame(all_monthly_corrs)
+            all_monthly_corrs_df = DataFrame(all_monthly_corrs)
 
             # Calculate average monthly correlations
             mean_corrs_df = (
@@ -116,7 +116,7 @@ class Persistence(Empirical):
         draw: bool = False,
         path: Optional[str] = None,
         decimal: Optional[int] = None,
-    ) -> pd.DataFrame:
+    ) -> DataFrame:
         """Calculate the transition matrix for a specified variable and lag.
 
         This method computes the transition matrix that shows how groups change over time based on
@@ -173,7 +173,7 @@ class Persistence(Empirical):
         )
 
         # Create DataFrame for the transition matrix
-        transition_matrix_df = pd.DataFrame(
+        transition_matrix_df = DataFrame(
             transition_matrix, columns=range(1, group + 1), index=range(1, group + 1)
         )
 

{anomalylab-0.2.8 → anomalylab-0.3.0}/anomalylab/empirical/portfolio.py

@@ -69,8 +69,8 @@ class PortfolioAnalysis(Empirical):
         self,
         vars: Union[str, list[str]],
         groups: Union[int, list[int]],
-        type: Optional[str] = None,
-    ) -> pd.DataFrame:
+        sort_type: Optional[str] = None,
+    ) -> DataFrame:
         """Group variables into portfolios based on specified groups.
 
         This method creates portfolios for the specified variables in the panel data.
@@ -78,7 +78,7 @@ class PortfolioAnalysis(Empirical):
         Args:
             vars (list of str): List of variables to group.
             groups (list of int): List of integers defining the number of groups for each variable.
-            type (str, optional): Type of grouping, can be 'dependent' to adjust based on the previous variable.
+            sort_type (str, optional): Type of sorting, can be 'dependent' to adjust based on the previous variable.
 
         Returns:
             DataFrame: A DataFrame with new columns for grouped variables.
@@ -105,7 +105,7 @@ class PortfolioAnalysis(Empirical):
         # Adjust group definitions
         group_col = [self.time]
         for i, var in enumerate(vars):
-            if type == "dependent" and i > 0:
+            if sort_type == "dependent" and i > 0:
                 group_col.append(f"{vars[i-1]}_g{groups[i-1]}")
                 # Grouping dependent on the previous variable
                 out_df[f"{var}_g{groups[i]}"] = (
@@ -142,7 +142,7 @@ class PortfolioAnalysis(Empirical):
 
         return out_df
 
-    def _claculate_value(self, df: pd.DataFrame, decimal: Optional[int] = None) -> dict:
+    def _claculate_value(self, df: DataFrame, decimal: Optional[int] = None) -> dict:
         """Calculate various portfolio performance metrics.
 
         This method computes mean returns, t-values, Sharpe ratios, and model-adjusted alpha and t values.
@@ -159,7 +159,7 @@ class PortfolioAnalysis(Empirical):
 
         return {**stat_dict, **factors_dict, **sharpe_dict}
 
-    def _calculate_mean_and_t_value(self, df: pd.DataFrame) -> dict:
+    def _calculate_mean_and_t_value(self, df: DataFrame) -> dict:
         """Calculate mean and t-value for the dependent variable.
 
         This method computes the mean return and its t-value assuming the null hypothesis
@@ -176,7 +176,7 @@ class PortfolioAnalysis(Empirical):
         lag = math.ceil(4 * (T / 100) ** (4 / 25))
 
         Y = df[self.endog].values
-        X = pd.DataFrame({"constant": [1] * len(df[self.endog])}).values
+        X = DataFrame({"constant": [1] * len(df[self.endog])}).values
         reg = sm.OLS(Y, X).fit(
             cov_type="HAC", cov_kwds={"maxlags": lag, "use_correction": False}
         )
@@ -190,7 +190,7 @@ class PortfolioAnalysis(Empirical):
 
         return stat_dict
 
-    def _calculate_alpha_and_t_value(self, df: pd.DataFrame) -> dict:
+    def _calculate_alpha_and_t_value(self, df: DataFrame) -> dict:
         """Calculate alpha and t-value for specified models.
 
         This method computes alpha values and their t-statistics for various regression models
@@ -239,7 +239,7 @@ class PortfolioAnalysis(Empirical):
         else:
             return {}
 
-    def _calculate_sharpe(self, df: pd.DataFrame, decimal: Optional[int] = 0) -> dict:
+    def _calculate_sharpe(self, df: DataFrame, decimal: Optional[int] = 0) -> dict:
         """Calculate the Sharpe ratio for the dependent variable.
 
         This method computes the annualized Sharpe ratio based on the mean and standard deviation
@@ -280,6 +280,7 @@ class PortfolioAnalysis(Empirical):
             core_g (int): The group number for portfolio grouping of the core variable.
             format (bool): Whether to format the output for display. Defaults to False.
             decimal (Optional[int]): The number of decimal places for formatting. Defaults to None.
+            factor_return (bool): Whether to output factor returns in the analysis. Defaults to False.
 
         Returns:
             tuple: A tuple containing the equal-weighted and value-weighted results DataFrames.
@@ -299,7 +300,7 @@ class PortfolioAnalysis(Empirical):
         )  # type: ignore
         vw_ret_d.index.names = [self.time, core_var]
 
-        def process_group(group: pd.DataFrame) -> pd.DataFrame:
+        def process_group(group: DataFrame) -> Series:
             """Process each group to calculate differences and prepare the output.
 
             This function computes the difference between the highest portfolio and the lowest
@@ -309,7 +310,7 @@ class PortfolioAnalysis(Empirical):
                 group (DataFrame): The grouped DataFrame for which to process data.
 
             Returns:
-                DataFrame: The processed DataFrame with differences and averages.
+                Series: The processed Series with differences and averages.
             """
             group = group.sort_index(axis=0, level=[0, 1])
 
@@ -318,7 +319,7 @@ class PortfolioAnalysis(Empirical):
                 [(group.index.get_level_values(0)[0], "Diff")],
                 names=[self.time, core_var],
             )
-            core_diff = pd.Series(core_diff, index=new_index)
+            core_diff = Series(core_diff, index=new_index)
 
             return pd.concat([group, core_diff])
 
@@ -358,7 +359,7 @@ class PortfolioAnalysis(Empirical):
 
         def calculate_time_series_metrics(
             series: Series, format: bool = format
-        ) -> pd.DataFrame:
+        ) -> DataFrame:
             """Calculate metrics for each time series and format results.
 
             This function computes performance metrics for each time series and formats the results
@@ -383,7 +384,7 @@ class PortfolioAnalysis(Empirical):
                 values[core_var] = key
                 data.append(values)
 
-            combined_results = pd.DataFrame(data)
+            combined_results = DataFrame(data)
 
             combined_results.set_index(core_var, inplace=True)
 
@@ -424,7 +425,7 @@ class PortfolioAnalysis(Empirical):
         core_g: int,
         pivot: bool = True,
         format: bool = False,
-        type: str = "dependent",
+        sort_type: str = "dependent",
         decimal: Optional[int] = None,
         factor_return: bool = False,
     ) -> tuple:
@@ -443,6 +444,7 @@ class PortfolioAnalysis(Empirical):
             format (bool): Whether to format the output for display. Defaults to False.
             type (str): Type of grouping, can be 'dependent' or 'independent'. Defaults to 'dependent'.
             decimal (Optional[int]): The number of decimal places to round to. Defaults to None.
+            factor_return (bool): Whether to output factor returns in the analysis. Defaults to False.
 
         Returns:
             tuple: A tuple containing the equal-weighted and value-weighted results DataFrames.
@@ -450,7 +452,7 @@ class PortfolioAnalysis(Empirical):
         data_d = self.GroupN(
             [sort_var, core_var],
             [sort_g, core_g],
-            type=type,
+            sort_type=sort_type,
         )
 
         ew_ret_d = data_d.groupby(
@@ -465,7 +467,7 @@ class PortfolioAnalysis(Empirical):
         )
         vw_ret_d.index.names = [self.time, sort_var, core_var]
 
-        def process_group(group: pd.DataFrame) -> pd.DataFrame:
+        def process_group(group: DataFrame) -> DataFrame:
             """Process each group to calculate differences and averages.
 
             This function computes the difference between the highest portfolio and lowest portfolio,
@@ -527,7 +529,7 @@ class PortfolioAnalysis(Empirical):
         if factor_return:
             return ew_ret_d, vw_ret_d
 
-        def generate_time_series_dict(df: pd.DataFrame) -> dict:
+        def generate_time_series_dict(df: DataFrame) -> dict:
             """Generate a dictionary of time series data from the DataFrame.
 
             This function extracts time series for each unique combination of sorting and core variables.
@@ -554,8 +556,8 @@ class PortfolioAnalysis(Empirical):
             return time_series_dict
 
         def calculate_time_series_metrics(
-            df: pd.DataFrame, pivot: bool = pivot, format: bool = format
-        ) -> pd.DataFrame:
+            df: DataFrame, pivot: bool = pivot, format: bool = format
+        ) -> DataFrame:
             """Calculate metrics for each time series and format results.
 
             This function computes performance metrics for each time series and formats the results
@@ -583,7 +585,7 @@ class PortfolioAnalysis(Empirical):
                 values[core_var] = key[1]
                 data.append(values)
 
-            combined_results = pd.DataFrame(data)
+            combined_results = DataFrame(data)
 
             combined_results.set_index([sort_var, core_var], inplace=True)
 
@@ -609,16 +611,16 @@ class PortfolioAnalysis(Empirical):
                 :, ~combined_results.columns.str.endswith("p")
             ]
 
-            def reorder_diff_avg(df: pd.DataFrame) -> pd.DataFrame:
+            def reorder_diff_avg(df: DataFrame) -> DataFrame:
                 """Reorder the rows and columns of a DataFrame to place 'Diff' before 'Avg'.
 
                 This function rearranges the DataFrame to improve readability.
 
                 Args:
-                    df (pd.DataFrame): The DataFrame to reorder.
+                    df (DataFrame): The DataFrame to reorder.
 
                 Returns:
-                    pd.DataFrame: The reordered DataFrame.
+                    DataFrame: The reordered DataFrame.
                 """
                 columns_order = [
                     col for col in df.columns if col not in ["Diff", "Avg"]

{anomalylab-0.2.8 → anomalylab-0.3.0}/anomalylab/preprocess/outliers.py

@@ -40,7 +40,7 @@ class OutlierMethod:
         Returns:
             Series: A new Series with winsorized values.
         """
-        return pd.Series(
+        return Series(
             data=np.where(
                 series.isnull(),
                 np.nan,
@@ -69,7 +69,7 @@ class OutlierMethod:
         Returns:
             Series: A new Series with truncated values.
         """
-        return pd.Series(
+        return Series(
             data=np.where(
                 series.isnull(),
                 np.nan,

{anomalylab-0.2.8 → anomalylab-0.3.0}/anomalylab/structure/data.py

@@ -21,18 +21,11 @@ class Data(ABC):
         1. Check if the columns are valid.
         2. Preprocess the data.
         3. Set the flag if needed.
-        4. Call the other_init method if needed.
         """
-        if self.name is None:
-            self.name = "anomaly"
+
         self._check_columns()
         self._preprocess()
         self.set_flag()
-        self.other_init()
-
-    def other_init(self) -> None:
-        """This method is a placeholder for additional initialization logic."""
-        pass
 
     def set_flag(self) -> None:
         """This method is meant to be overridden by subclasses to set flags."""

{anomalylab-0.2.8 → anomalylab-0.3.0}/anomalylab/structure/panel_data.py

@@ -20,7 +20,7 @@ class PanelData(Data):
         frequency (Literal["D", "M", "Y"]):
             The frequency of the data. Defaults to "M".
         ret (str):
-            The column name for the excess return. Defaults to "return".
+            The column name for the excess return. Defaults to None.
         classifications (list[str]):
             The list of classification columns.
         drop_all_chars_missing (bool):
@@ -30,9 +30,10 @@ class PanelData(Data):
     id: str = "permno"
     time: str = "date"
     frequency: Literal["D", "M", "Y"] = "M"
-    ret: str = "return"
+    ret: Optional[str] = None
     classifications: Optional[list[str] | str] = None
     drop_all_chars_missing: bool = False
+    is_copy: bool = False
 
     def set_flag(self) -> None:
         """Set default flags for the `PanelData` object."""
@@ -57,10 +58,12 @@ class PanelData(Data):
 
         This method identifies remaining columns as firm characteristics, excluding classifications.
         """
+        if self.is_copy:
+            self.df = copy.deepcopy(self.df)
         self.df[self.id] = self.df[self.id].astype(int)
         self.df[self.time] = pd.to_datetime(self.df[self.time], format="ISO8601")
         self.df[self.time] = self.df[self.time].dt.to_period(freq=self.frequency)
-        self.df = self.df.sort_values(by=[self.time, self.id])
+        self.df.sort_values(by=[self.time, self.id], inplace=True)
         basic_column = (
             [self.id, self.time] if self.ret is None else [self.id, self.time, self.ret]
         )
@@ -208,6 +211,7 @@ if __name__ == "__main__":
         ret="return",
         classifications="industry",
         drop_all_chars_missing=True,
+        is_copy=False,
     )
     pp(panel_data)
     pp(panel_data.df)

{anomalylab-0.2.8 → anomalylab-0.3.0}/anomalylab/structure/time_series.py

@@ -24,6 +24,7 @@ class TimeSeries(Data):
     time: str = "date"
     frequency: Literal["D", "M", "Y"] = "M"
     factors: list[str] = field(init=False)
+    is_copy: bool = False
 
     def __repr__(self) -> str:
         return f"TimeSeriesData({self.name})"  # todo: add frequency
@@ -34,6 +35,8 @@ class TimeSeries(Data):
 
         This method renames the time column to a standardized name and identifies remaining columns as factors.
         """
+        if self.is_copy:
+            self.df = copy.deepcopy(self.df)
         self.df[self.time] = pd.to_datetime(self.df[self.time], format="ISO8601")
         self.df[self.time] = self.df[self.time].dt.to_period(freq=self.frequency)
         self.df = self.df.sort_values(by=self.time)

{anomalylab-0.2.8 → anomalylab-0.3.0}/setup.py

@@ -2,7 +2,7 @@ from setuptools import find_packages, setup
 
 setup(
     name="AnomalyLab",
-    version="0.2.8",
+    version="0.3.0",
     author="FinPhd",
     # author_email="your.email@example.com",
     description="A Python package for empirical asset pricing analysis.",