gitlabds 1.1.0__tar.gz → 1.1.2__tar.gz

{gitlabds-1.1.0 → gitlabds-1.1.2}/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: gitlabds
-Version: 1.1.0
+Version: 1.1.2
 Summary: Gitlab Data Science and Modeling Tools
 Home-page: https://gitlab.com/gitlab-data/gitlabds
 Author: Kevin Dietz

{gitlabds-1.1.0 → gitlabds-1.1.2}/gitlabds/dummies.py

@@ -43,6 +43,10 @@ def dummy_code(
     DataFrame with dummy-coded columns. Categorical columns that were dummy coded will be dropped from the dataframe.
     """
     
+
+
+
+
     if columns == "all":
         var_list = df.columns.tolist()
         print("Will examine all variables as candidate for dummy coding")
@@ -205,9 +209,11 @@ def dummy_top(
             
         levels = levels.loc[levels > min_threshold].index.tolist()
 
+        dummy_dfs = []        
         for l in levels:
-            dummy_name = str(d) + "_dummy_" + str(l)
-            new_df[dummy_name] = np.where(new_df[d] == str(l), 1, 0)
+            dummy_name = f"{d}_dummy_{l}"
+            dummy_series = pd.Series(np.where(new_df[d] == str(1), 1,0), name=dummy_name, index=new_df.index)
+            dummy_dfs.append(dummy_series)
             
             if output_file != None: 
                 
@@ -217,6 +223,8 @@ def dummy_top(
                 else:    
                     f.write(f"\n    df['{dummy_name}'] = np.where(df['{d}'] == '{l}', 1, 0)")
 
+        new_df = pd.concat([new_df] + dummy_dfs, axis=1)
+
         # drop categorical fields after they have been dummy coded
         if drop_categorial == True:
             new_df.drop(columns=[d], inplace=True)
@@ -225,4 +233,4 @@ def dummy_top(
         f.write('\n\n    return df')
         f.close()
 
-    return new_df
+    return new_df

{gitlabds-1.1.0 → gitlabds-1.1.2}/gitlabds/outliers.py

@@ -112,7 +112,7 @@ def mad_outliers(
             # Windsor values if they fall below the windsor_threshold
             if ((min_aff / len(df) < windsor_threshold) and (min_aff > 0) and (new_min != old_min)):
                 
-                df[v].clip(lower=new_min, inplace=True)
+                df[v] = df[v].clip(lower=new_min)
 
                 if verbose == True:
                     print(f"{v} - MAD: {mad}; Lower Windsor Value: {new_min}, Rows affected {min_aff} ({min_aff/len(df)*100})")
@@ -122,7 +122,7 @@ def mad_outliers(
 
             if ((max_aff / len(df) < windsor_threshold) and (max_aff > 0) and (new_max != old_max)):
                 
-                df[v].clip(upper=new_max, inplace=True)
+                df[v] = df[v].clip(upper=new_max)
 
                 if verbose == True:
                     print(f"{v} - MAD: {mad}; Upper Windsor Value: {new_max}, Rows affected {max_aff} ({max_aff/len(df)*100}%)")
@@ -179,7 +179,7 @@ def mad_outliers(
             # Windsor values if they fall below the windsor_threshold
             if ((min_aff / len(df) < windsor_threshold) and (min_aff > 0) and (new_min != old_min)):
                 
-                new_df[v].clip(lower=new_min, inplace=True)
+                new_df[v] = new_df[v].clip(lower=new_min)
 
                 if verbose == True:
                     print(f"{v} - MAD: {mad}; Lower Windsor Value: {new_min}, Rows affected {min_aff} ({min_aff/len(df)*100})")
@@ -189,7 +189,7 @@ def mad_outliers(
 
             if ((max_aff / len(df) < windsor_threshold) and (max_aff > 0) and (new_max != old_max)):
                 
-                new_df[v].clip(upper=new_max, inplace=True)
+                new_df[v] = new_df[v].clip(upper=new_max)
 
                 if verbose == True:
                     print(f"{v} - MAD: {mad}; Upper Windsor Value: {new_max}, Rows affected {max_aff} ({max_aff/len(df)*100}%)")

{gitlabds-1.1.0 → gitlabds-1.1.2}/gitlabds/split_data.py

@@ -51,9 +51,12 @@ def split_data(df:pd.DataFrame, train_pct:float=0.7, dv:str=None, dv_threshold:f
 
             print(f'Outcome variable "{dv}" pct: {dv_pct}. Below the dv_threshold value of {dv_threshold}. Will up-sample with SMOTE-NC...')
 
-            # Get list of binary variables (mostly categorical dummy codes)
-            cats = pd.DataFrame(x_train.select_dtypes(include=["number"]).nunique(dropna=False, axis=0))
-            cats = np.where(cats[0] == 2, True, False)
+            # Get numeric columns
+            numeric_cols = x_train.select_dtypes(include=["number"]).columns
+            
+            # Create categorical feature indices list. Should just be dummy features that have 2 values
+            cats = [i for i, col in enumerate(numeric_cols) 
+                   if x_train[col].nunique(dropna=False) == 2]
 
             # SMOTENC
             sm = SMOTENC(random_state=random_state,

{gitlabds-1.1.0 → gitlabds-1.1.2}/gitlabds.egg-info/PKG-INFO

@@ -1,6 +1,6 @@
 Metadata-Version: 2.1
 Name: gitlabds
-Version: 1.1.0
+Version: 1.1.2
 Summary: Gitlab Data Science and Modeling Tools
 Home-page: https://gitlab.com/gitlab-data/gitlabds
 Author: Kevin Dietz