Added decoding function to covert back to normal values

src/modules/generationtools/categorical.py

@@ -76,3 +76,22 @@ def categorical_convert(col):
     # sample from the distributions and return that value
     return col.apply(lambda x: distributions[x].rvs()), limits
 
+
+def undo_cat(col, lim):
+    """Convert the categorical column to normalized valus
+    
+    Arguments:
+        col {Column} -- The dataframe's column
+        lim {Dict} -- The dictionary containing the limits per catergorical column
+    
+    Returns:
+        Column -- The new column
+    """
+
+
+    def cat_decode(x, limits):
+        for k, v in limits.items():
+            if x < k:
+                return v
+
+    return col.apply(lambda x: cat_decode(x, lim))

src/modules/generationtools/synthesize.py

@@ -9,7 +9,7 @@
 import numpy as np
 from scipy.stats import expon, truncnorm, beta, uniform, norm
 from .cleandata import MissingValues, DatetimeToEPOCH
-from .categorical import identify, categorical_convert
+from .categorical import identify, categorical_convert, undo_cat
 
 def sample(f, sigma):
     '''
@@ -57,15 +57,13 @@ def synthesize_table(file_in, file_out, lines = 0):
     df = MissingValues(df)
     df = DatetimeToEPOCH(df)
 
+    cat_cols = {}
     limits = {}
-    counter = 0
     for col in df:
         if(identify(df[col])):
             new_col, limit = categorical_convert(df[col])
-            limits[counter] = limit
-            counter += 1
+            limits[col] = limit
             df[col] = new_col
-
     # calculate distributions and covariances using tools in model_generation.py
     dists, pvalues, params = mg.findBestDistribution(df)
     f = (dists, params)
@@ -78,8 +76,9 @@ def synthesize_table(file_in, file_out, lines = 0):
     new_df = pd.DataFrame(columns = list(df))
     for k in range(lines):
         new_df.loc[k] = sample(f, sigma)
-    new_df.to_csv(file_out, index = False)
 
+    for col in new_df.columns:
+        if col in limits:
+            new_df[col] = undo_cat(new_df[col], limits[col])
 
-if __name__ == "__main__":
-    synthesize_table("test.csv", "s_test.csv")
+    new_df.to_csv(file_out, index = False)