econmethods 3.0.4__tar.gz → 3.0.5.1__tar.gz

{econmethods-3.0.4 → econmethods-3.0.5.1}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: econmethods
-Version: 3.0.4
+Version: 3.0.5.1
 Summary: A python package implementing various econometrical tests and estimators
 Home-page: https://github.com/NaturionBG/econmethods
 Author: NaturionBG

{econmethods-3.0.4 → econmethods-3.0.5.1}/econmethods/Lib.py

@@ -412,13 +412,14 @@ class FECM:
     return means
   
   def build_GLS(self, w_err: float) -> np.ndarray:
-    re = self.__lr_df.copy(deep=True)
-    sigma2 = np.sum((sm.OLS(re.iloc[:, 2], sm.add_constant(re.iloc[:, 3:])).fit()).resid**2) / (self.__N*self.__T - self.__exog)
-    sigma_u = sigma2 - w_err
+    re = self.__lr_df.groupby('region')[self.__lr_df.columns[2:]].mean()
+    temp = sm.OLS(re.iloc[:, 2], sm.add_constant(re.iloc[:, 3:])).fit()
+    sigma2 = temp.ssr / temp.df_resid
+    sigma_u = sigma2 - w_err/self.__T
     if sigma_u <= 0:
       sigma_u = 0
     sig = np.full((self.__T, self.__T), sigma_u)
-    np.fill_diagonal(sig, sigma2)
+    np.fill_diagonal(sig, sigma_u+w_err)
     matrix = np.kron(np.eye(self.__N), sig)
     return matrix
   
@@ -460,11 +461,11 @@ class FECM:
     else:
       lr_fe = self.build_FE()
       if self.__lr_c:
-        resid = np.sum(sm.OLS(lr_fe.loc[:, 'target'], sm.add_constant(lr_fe.iloc[:, 3:])).fit().resid**2) / (self.__N*(self.__T - 1) - self.__exog)
+        resid = sm.OLS(lr_fe.loc[:, 'target'], sm.add_constant(lr_fe.iloc[:, 3:])).fit().mse_resid
         lr_re_matrix = self.build_GLS(resid)
         return sm.GLS(self.__lr_df.loc[:, 'target'], sm.add_constant(self.__lr_df.iloc[:, 3:]), lr_re_matrix).fit()
       else:
-        resid = np.sum(sm.OLS(lr_fe.loc[:, 'target'], lr_fe.iloc[:, 3:]).fit().resid**2) / (self.__N*(self.__T - 1) - self.__exog)
+        resid = sm.OLS(lr_fe.loc[:, 'target'], lr_fe.iloc[:, 3:]).fit().mse_resid
         lr_re_matrix = self.build_GLS(resid)
         return sm.GLS(self.__lr_df.loc[:, 'target'], self.__lr_df.iloc[:, 3:], lr_re_matrix).fit()
   
@@ -515,7 +516,7 @@ class FECM:
           model = model.drop(columns=drop)
         else:
           self.__C = False
-          drop = drop.remove('const')
+          drop.remove('const')
         if self.__C:
           est.append(sm.OLS(model['target_diff'], sm.add_constant(model.iloc[:, 3:])).fit())
         else:
@@ -533,7 +534,7 @@ class FECM:
           model = model.drop(columns=drop)
         else:
           self.__C = False
-          drop = drop.remove('const')
+          drop.remove('const')
         if self.__C:
           est.append(sm.OLS(model['target_diff'], sm.add_constant(model.iloc[:, 3:])).fit())
         else:
@@ -542,7 +543,7 @@ class FECM:
           self.__mg_units[i] = self.__mg_units[i].drop(columns=drop)
         else:
           self.__C = False
-          drop = drop.remove('const')
+          drop.remove('const')
       return est
     elif self.__method == 'ccep':
       units = self.get_ccemg_frames(self.__lag)
@@ -701,41 +702,31 @@ class CDTwoWay:
       print(f'p-value = {pval} < alpha = {self.__alpha}\n There is Significant Cross-Sectional Dependence in your data according to the CD-test. \n Significance level: {self.__alpha*100}%')
     else:
       print(f'p-value = {pval} > alpha = {self.__alpha} There is No Significant Cross-Sectional Dependence in your data according to the CD-test. \n Significance level: {self.__alpha*100}%')
-
-
 class SlopeHomogeneityF:
   '''
-  <h2>The implementation of the F-test for data slope Homogeneity in panel data.</h2>
-  <hr>
-  <h2>Hypotheses:</h2>
-  <hr>
-  <ul>
-  <li> <em>H0</em>: b_i = b_j given that i != j</li>
-  <li><em>H_1</em>: b_i != b_j for at least one such pair of (i, j) that i != j </li>
-  </ul>
-  <hr>
-  <h2>PARAMETERS:</h2>
-  <hr>
-  <ul>
-  <li> <strong>df</strong>: a Pandas DataFrame containing panel data. Make sure your data is structured in this exact order (by column index):
-    <ul>
-    <li>0 - Spatial units. The data must be homogenous, e.g. only cities, contries, regions, etc.</li>
-    <li>1 - Temporal units. The data must be homogenous, e.g. only years, months, quarters, etc.</li>
-    <li>2 - Your target/endogenous variable. The data must not contain NaN values.</li>
-    <li>3+ - Your exogenous variables. The data must not contain NaN values.</li>
-    </ul>
-  <li> <strong>level</strong>: Your significance level to test at. Defaults to 5%.</li>
-  <li> <strong>intercept</strong>: Specify whether your model has an intercept. Defaults to True.</li>
-  <li> <strong>skip_premise</strong>: Set to True if you wish to ignore heteroskedasticity. Not a recommended option, parameter defaults to False</li>
-  </ul>
-  <hr>
-    <h3><em>Note that this test has got certain assumptions, such as error homoskedasticity, N < T.
-     The <FTwoWay> instance will check for these conditions to be met, and will yield an error if they are not.</em></h3>
-  <hr>
-  <h2>RETURNS:</h2>
-  <ol>
-  <li>Via the instance.verdict() method - prints the result of the F-test.</li>
-  </ol>
+  The implementation of the F-test for data slope Homogeneity in panel data.
+  ---
+  Hypotheses:
+  ---
+  + H0: b_i = b_j given that i != j
+  + H_1: b_i != b_j for at least one such pair of (i, j) that i != j
+  ---
+  PARAMETERS:
+  ---
+  - **df:** a Pandas DataFrame containing panel data. Make sure your data is structured in this exact order (by column index):
+    + 0 - Spatial units. The data must be homogenous, e.g. only cities, contries, regions, etc.
+    + 1 - Temporal units. The data must be homogenous, e.g. only years, months, quarters, etc.
+    + 2 - Your target/endogenous variable. The data must not contain NaN values.
+    + 3+ - Your exogenous variables. The data must not contain NaN values.
+  - **level:** Your significance level to test at. Defaults to 5%.
+  - **intercept:** Specify whether your model has an intercept. Defaults to True.
+  - **skip_premise:** Set to True if you wish to ignore heteroskedasticity. Not a recommended option, parameter defaults to False
+    *Note that this test has got certain assumptions, such as error homoskedasticity, N < T.
+     The <FTwoWay> instance will check for these conditions to be met, and will yield an error if they are not.*
+  ---
+  **RETURNS:**
+
+  + Via the instance.verdict() method - prints the result of the F-test.
   '''
   def __init__(self, df: pd.DataFrame, level: int = 5, intercept: bool = True, skip_premise: bool = False) -> None:
     self.__df = df
@@ -787,4 +778,4 @@ class SlopeHomogeneityF:
     
   def __del__(self) -> None:
     pass
-  
+

{econmethods-3.0.4 → econmethods-3.0.5.1}/econmethods.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.4
 Name: econmethods
-Version: 3.0.4
+Version: 3.0.5.1
 Summary: A python package implementing various econometrical tests and estimators
 Home-page: https://github.com/NaturionBG/econmethods
 Author: NaturionBG

{econmethods-3.0.4 → econmethods-3.0.5.1}/setup.py

@@ -5,7 +5,7 @@ with open('README.md', 'r') as f:
   
 setup(
   name = 'econmethods',
-  version = 'v3.0.4',
+  version = 'v3.0.5.1',
   description='A python package implementing various econometrical tests and estimators',
   packages = find_packages(),
   long_description=long_desc,