synthesised categorical variable without encoding for bar dataset works if we don't include sex

Author: KoomenErina

main.py

@@ -2,11 +2,28 @@
 import pandas as pd
 import pyreadr
 
+def synBar():
+    df = pd.read_csv("bar_pass_prediction.csv")[[ 'race1', 'ugpa', 'bar']]
+    print(df.dtypes)
+    dtype_map = { 'race1': 'category', 'ugpa': 'float', 'bar': 'category'}
+
+    for (k,v) in dtype_map.items():
+
+        if v == 'category':
+            df = df.astype({k : "category"})
+
+    print(df.dtypes)
+    spop = Synthpop()
+    spop.fit(df,dtype_map)
+
+    synth_df = spop.generate(len(df))
+
+    print(synth_df.head())
 
 def synSD2011():
     df0 = pyreadr.read_r("SD2011.rda")['SD2011']
     #pd.read_csv("bar_pass_prediction.csv")
-    print(df0)
+    print(df0.dtypes)
     df = df0[['age', 'unempdur', 'income', 'sex']]#df0[['sex', 'race1', 'ugpa', 'bar']]
     #df.to_excel("inputData.xlsx")
     dtype_map ={
@@ -27,4 +44,7 @@ def synSD2011():
 
     synth_df = spop.generate(len(df))
 
-    print(synth_df.head())
+    print(synth_df.head())
+
+
+synBar()

synthpop/processor/processor.py

@@ -57,7 +57,7 @@ def preprocess(self, df, dtypes):
                                                               'nan_value': col_nan_category
                                                               }
 
-                        df[col].cat.add_categories(col_nan_category, inplace=True)
+                        df[col] = df[col].cat.add_categories(col_nan_category) #argument 'inplace' is deprecated and removed
                         df[col].fillna(col_nan_category, inplace=True)
 
                 # NaNs in numerical columns
@@ -79,7 +79,7 @@ def preprocess(self, df, dtypes):
                             df.loc[bool_series, col_nan_name] = cat_index
                         df.loc[col_all_nan_indices, col] = 0
 
-                        df[col_nan_name] = df[col_nan_name].astype('category')
+                        df.loc[:,col_nan_name] = df[col_nan_name].astype('category')
                         self.spop.df_dtypes[col_nan_name] = 'category'
 
         return df

synthpop/synthpop.py

@@ -57,6 +57,7 @@ def fit(self, df, dtypes=None):
         self.validator.check_processor()
         # preprocess
         processed_df = self.processor.preprocess(df, self.df_dtypes)
+        print(processed_df)
         self.processed_df_columns = processed_df.columns.tolist()
         self.n_processed_df_columns = len(self.processed_df_columns)