AnomalyLab 0.3.3__tar.gz → 0.3.4__tar.gz

{anomalylab-0.3.3 → anomalylab-0.3.4/AnomalyLab.egg-info}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: AnomalyLab
-Version: 0.3.3
+Version: 0.3.4
 Summary: A Python package for empirical asset pricing analysis.
 Author: FinPhd
 Classifier: Programming Language :: Python :: 3

{anomalylab-0.3.3 → anomalylab-0.3.4}/AnomalyLab.egg-info/SOURCES.txt

@@ -23,7 +23,6 @@ anomalylab/datasets/transition_matrix.png
 anomalylab/empirical/__init__.py
 anomalylab/empirical/correlation.py
 anomalylab/empirical/empirical.py
-anomalylab/empirical/factor_return.py
 anomalylab/empirical/fm_regression.py
 anomalylab/empirical/persistence.py
 anomalylab/empirical/portfolio.py

{anomalylab-0.3.3/AnomalyLab.egg-info → anomalylab-0.3.4}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: AnomalyLab
-Version: 0.3.3
+Version: 0.3.4
 Summary: A Python package for empirical asset pricing analysis.
 Author: FinPhd
 Classifier: Programming Language :: Python :: 3

{anomalylab-0.3.3 → anomalylab-0.3.4}/anomalylab/core/core.py

@@ -335,6 +335,7 @@ class Panel:
         industry_weighed_method: Literal["value", "equal"] = "value",
         is_winsorize: bool = False,
         is_normalize: bool = False,
+        dummy_no_norm: Optional[list[str] | str] = None,
         decimal: Optional[int] = None,
         return_intermediate: bool = False,
     ) -> DataFrame:
@@ -349,6 +350,7 @@ class Panel:
             industry_weighed_method=industry_weighed_method,
             is_winsorize=is_winsorize,
             is_normalize=is_normalize,
+            dummy_no_norm=dummy_no_norm,
             decimal=decimal,
             return_intermediate=return_intermediate,
         )

{anomalylab-0.3.3 → anomalylab-0.3.4}/anomalylab/empirical/correlation.py

@@ -85,6 +85,8 @@ if __name__ == "__main__":
 
     df: DataFrame = DataSet.get_panel_data()
 
-    panel: PanelData = PanelData(df=df, name="panel", classifications="industry")
+    panel: PanelData = PanelData(
+        df=df, name="Stocks", ret="return", classifications="industry"
+    )
     correlation: Correlation = Correlation(panel_data=panel)
     pp(correlation.average_correlation())

{anomalylab-0.3.3 → anomalylab-0.3.4}/anomalylab/empirical/fm_regression.py

@@ -92,6 +92,7 @@ class FamaMacBethRegression(Empirical):
         df: DataFrame,
         reg: RegModel,
         is_normalize: bool,
+        dummy_no_norm: list[str] = [],
         return_intermediate: bool = False,
     ) -> RegResult:
         """Performs Fama-MacBeth regression on the provided DataFrame.
@@ -104,6 +105,11 @@ class FamaMacBethRegression(Empirical):
             df (DataFrame): DataFrame containing the data for regression.
             reg (RegModel): Model specification containing endogenous and exogenous variables.
             is_normalize (bool): Indicates whether to normalize the exogenous variables.
+            dummy_no_norm (list[str]): List of variable names that should be excluded from normalization.
+                Typically, these are dummy variables that take values like 0 or 1,
+                where normalization may not make sense.
+            return_intermediate (bool): If True, returns intermediate regression results
+                (e.g., coefficients, t-values, and R²) for each time period.
 
         Returns:
             RegResult: Results of the regression including parameters, t-values, p-values, and adjusted R².
@@ -115,9 +121,27 @@ class FamaMacBethRegression(Empirical):
         df = df.groupby(self.time).filter(lambda x: len(x) > 1)
         lag: int = math.ceil(4 * (df[self.time].nunique() / 100) ** (4 / 25))
         if is_normalize:
-            df[exogenous] = df.groupby(self.time)[exogenous].transform(
-                func=lambda x: (x - x.mean()) / x.std()
-            )
+            dummy_no_norm = [col for col in dummy_no_norm if col in exogenous]
+
+            variables_to_normalize = [
+                col for col in exogenous if col not in dummy_no_norm
+            ]
+
+            for col in variables_to_normalize:
+                grouped_std = df.groupby(self.time)[col].std()
+
+                if (grouped_std == 0).any():
+                    problematic_group = grouped_std[grouped_std == 0].index.tolist()
+                    raise ValueError(
+                        f"Standard deviation is 0 for variable '{col}' in group(s) {problematic_group}. "
+                        f"Cannot normalize this variable. If '{col}' is a dummy variable, please add it to `dummy_no_norm`."
+                    )
+
+            if variables_to_normalize:
+                df[variables_to_normalize] = df.groupby(self.time)[
+                    variables_to_normalize
+                ].transform(func=lambda x: (x - x.mean()) / x.std())
+
         df[self.time] = df[self.time].dt.to_timestamp()
         df = df.set_index([self.id, self.time])
 
@@ -277,6 +301,7 @@ class FamaMacBethRegression(Empirical):
         industry_weighed_method: Literal["value", "equal"] = "value",
         is_winsorize: bool = False,
         is_normalize: bool = False,
+        dummy_no_norm: Optional[list[str] | str] = None,
         decimal: Optional[int] = None,
         return_intermediate: bool = False,  # New parameter to control whether intermediate results are returned
     ) -> DataFrame:
@@ -297,6 +322,8 @@ class FamaMacBethRegression(Empirical):
             industry_weighed_method (Literal["value", "equal"]): Method for weighting industries.
             is_winsorize (bool): Indicates whether to apply winsorization.
             is_normalize (bool): Indicates whether to normalize exogenous variables.
+            dummy_no_norm (Optional[list[str] | str]): Name(s) of dummy variables (e.g., 0 or 1)
+                that should be excluded from normalization.
             decimal (Optional[int]): Number of decimal places for rounding in output.
             return_intermediate (bool): If True, returns the intermediate results (e.g., coefficients for each time period).
 
@@ -307,6 +334,7 @@ class FamaMacBethRegression(Empirical):
         reg_models: RegModels = self._model_parse(
             regs=regs, endog=endog, exog=columns_to_list(exog)
         )
+        dummy_no_norm = columns_to_list(dummy_no_norm)
         self._winsorize(is_winsorize=is_winsorize, exog=reg_models.exogenous)
         self._industry_weighted(
             endog=reg_models.dependent,
@@ -361,8 +389,9 @@ if __name__ == "__main__":
 
     df: DataFrame = DataSet.get_panel_data()
 
-    panel: PanelData = PanelData(df=df, name="Stocks", classifications="industry")
-
+    panel: PanelData = PanelData(
+        df=df, name="Stocks", ret="return", classifications="industry"
+    )
     fm = FamaMacBethRegression(panel_data=panel)
     result = fm.fit(
         # endog="return",
@@ -386,8 +415,8 @@ if __name__ == "__main__":
         # weight="MktCap",
         is_winsorize=True,
         is_normalize=True,
-        return_intermediate=True,
+        return_intermediate=False,
         # decimal=2,
     )
     pp(result)
-    pp(result[0])
+    # pp(result[0])

{anomalylab-0.3.3 → anomalylab-0.3.4}/anomalylab/empirical/persistence.py

@@ -204,8 +204,9 @@ if __name__ == "__main__":
 
     df: DataFrame = DataSet.get_panel_data()
 
-    panel: PanelData = PanelData(df=df, name="panel", classifications="industry")
-
+    panel: PanelData = PanelData(
+        df=df, name="Stocks", ret="return", classifications="industry"
+    )
     persistence = Persistence(panel)
     pp(persistence.average_persistence(periods=[1, 3, 6, 12, 36, 60]))
     pp(

{anomalylab-0.3.3 → anomalylab-0.3.4}/anomalylab/empirical/portfolio.py

@@ -680,7 +680,9 @@ if __name__ == "__main__":
         "FF5": ["MKT(5F)", "SMB(5F)", "HML(5F)", "RMW(5F)", "CMA(5F)"],
     }
 
-    panel: PanelData = PanelData(df=df, name="Stocks", classifications="industry")
+    panel: PanelData = PanelData(
+        df=df, name="Stocks", ret="return", classifications="industry"
+    )
     time_series: TimeSeries = TimeSeries(df=ts, name="Factor Series")
 
     portfolio = PortfolioAnalysis(

{anomalylab-0.3.3 → anomalylab-0.3.4}/anomalylab/empirical/summary.py

@@ -125,7 +125,9 @@ if __name__ == "__main__":
 
     df: DataFrame = DataSet.get_panel_data()
 
-    panel: PanelData = PanelData(df=df, name="panel", classifications="industry")
+    panel: PanelData = PanelData(
+        df=df, name="Stocks", ret="return", classifications="industry"
+    )
     summary = Summary(panel_data=panel)
     pp(
         summary.average_statistics(

{anomalylab-0.3.3 → anomalylab-0.3.4}/anomalylab/preprocess/normalize.py

@@ -157,7 +157,9 @@ if __name__ == "__main__":
 
     df: DataFrame = DataSet.get_panel_data()
 
-    panel: PanelData = PanelData(df=df, name="Stocks", classifications="industry")
+    panel: PanelData = PanelData(
+        df=df, name="Stocks", ret="return", classifications="industry"
+    )
     norm: Normalize = Normalize(panel_data=panel)
     norm.normalize(
         # columns="MktCap",

{anomalylab-0.3.3 → anomalylab-0.3.4}/anomalylab/preprocess/outliers.py

@@ -193,8 +193,9 @@ if __name__ == "__main__":
 
     df: DataFrame = DataSet.get_panel_data()
 
-    panel: PanelData = PanelData(df=df, name="Stocks", classifications="industry")
-
+    panel: PanelData = PanelData(
+        df=df, name="Stocks", ret="return", classifications="industry"
+    )
     winsorize = OutlierHandler(panel_data=panel)
     winsorize.winsorize(
         # columns="MktCap",

{anomalylab-0.3.3 → anomalylab-0.3.4}/anomalylab/preprocess/shift.py

@@ -110,7 +110,9 @@ if __name__ == "__main__":
 
     df: DataFrame = DataSet.get_panel_data()
 
-    panel: PanelData = PanelData(df=df, name="Stocks", classifications="industry")
+    panel: PanelData = PanelData(
+        df=df, name="Stocks", ret="return", classifications="industry"
+    )
     shift = Shift(panel_data=panel)
     shift.shift(
         # columns="MktCap",

{anomalylab-0.3.3 → anomalylab-0.3.4}/setup.py

@@ -2,7 +2,7 @@ from setuptools import find_packages, setup
 
 setup(
     name="AnomalyLab",
-    version="0.3.3",
+    version="0.3.4",
     author="FinPhd",
     # author_email="your.email@example.com",
     description="A Python package for empirical asset pricing analysis.",

anomalylab-0.3.3/anomalylab/empirical/factor_return.py

@@ -1,43 +0,0 @@
-from pandas.core.frame import DataFrame
-
-from anomalylab.config import *
-from anomalylab.empirical.empirical import Empirical
-from anomalylab.empirical.portfolio import PortfolioAnalysis
-from anomalylab.structure import PanelData, TimeSeries
-from anomalylab.utils.imports import *
-from anomalylab.utils.utils import *
-
-if __name__ == "__main__":
-    from anomalylab.datasets import DataSet
-
-    df: DataFrame = DataSet.get_panel_data()
-    ts: DataFrame = DataSet.get_time_series_data()
-    Models: dict[str, list[str]] = {
-        "CAPM": ["MKT(3F)"],
-        "FF3": ["MKT(3F)", "SMB(3F)", "HML(3F)"],
-        "FF5": ["MKT(5F)", "SMB(5F)", "HML(5F)", "RMW(5F)", "CMA(5F)"],
-    }
-
-    panel: PanelData = PanelData(df=df, name="Stocks", classifications="industry")
-    time_series: TimeSeries = TimeSeries(df=ts, name="Factor Series")
-
-    portfolio = PortfolioAnalysis(
-        panel,
-        endog="return",
-        weight="MktCap",
-        # models=Models,
-        # factors_series=time_series,
-    )
-
-    group = portfolio.GroupN(["MktCap", "Illiq", "IdioVol"], [3, 3, 3])
-    pp(group)
-
-    # uni_ew, uni_vw = portfolio.univariate_analysis("Illiq", 10)
-    # pp(uni_ew)
-    # pp(uni_vw)
-
-    # bi_ew, bi_vw = portfolio.bivariate_analysis(
-    #     "Illiq", "IdioVol", 10, 10, True, False, "dependent"
-    # )
-    # pp(bi_ew)
-    # pp(bi_vw)