@@ -45,25 +45,49 @@ def clean_df_generic(df,threshold):
     print(df)
     return df
 
-   
 
-        
+
+
+def merge_columns(df,column_name1,column_name2):
+    print(df[[column_name1,column_name2]])
+    for index, row in df.iterrows():
+        if row[column_name1] != row[column_name2]:        
+            if pd.isnull(row[column_name1])==True:
+                #print(row[column_name1]," ",row[column_name2])
+                df.at[index,column_name1] = row[column_name2]
+            if pd.isnull(row[column_name2])==True:
+                #print(row[column_name1]," ",row[column_name2])
+                df.at[index,column_name2] = row[column_name1]
+    print(df[[column_name1,column_name2]])
+    df = df.drop(columns=[column_name1])
+    df = df.dropna(subset=[column_name2])
+    print(df)
+    return df
 
 
 ##generic cleaning 
 
+#order 
 df = read_data.load_dataframe("order.pkl")
-delete_columns_with_nullonly_or_morethan99p(df)
 df = clean_df_generic(df,1)
 
-
-df2 = read_data.load_dataframe("click.pkl")
-df2 = clean_df_generic(df2,1)
+#clickstream
+#df2 = read_data.load_dataframe("click.pkl")
+#df2 = clean_df_generic(df2,1)
 
 
 
 ##semiauto_cleaning 
 
+#order
+df = merge_columns(df,"Registration Gender","Gender")
+
+
+
+# merge columns Registration Gender and Gender to Gender and drop Registration Gender
+
+    
+
 
 
 
 
@@ -17,7 +17,7 @@ def read_raw_data(filename,column_file,samplename,dataset_name):
     # see for example column DoYouPurchaseForOthers before and after execution
     #df = recreate_dtypes(df, dtypes)
     ### end of deactivation 
-    create_sampel_excel(df,samplename)
+    #create_sampel_excel(df,samplename)
     save_dataframe(df,dataset_name)
     print(df)
     return df
@@ -108,7 +108,7 @@ def create_sampel_excel(dataframe,samplename):
     xlsx_writer.save()
 
 def create_csv_from_dataframe(dataframe,filename):
-    filename = "../00_raw_data/"+fiename
+    filename = "../00_raw_data/"+filename
     path_to_file = relative_to_absolute(filename)
     df = dataframe
     df.to_csv(path_or_buf=path_to_file,index=False)