fixed kg_maker

Closes #20 (closed)

src/kg_maker.py

@@ -5,6 +5,7 @@ import networkx as nx
 import pandas as pd
 import networkx as nx
 import matplotlib.pyplot as plt
+from collections import Counter
 
 
 class KgMaker:
@@ -56,23 +57,25 @@ class KgMaker:
     
     def group_meta_data(self)->pd.DataFrame:
         """Groups meta data by activity
-
-        Returns:
-            pd.DataFrame
+        returns: Dataframe
         """
-        columns_all=self.df.columns
-        meta_data=[]
-        for col in columns_all:
-            if col not in ["timestamp","case id","activity"]:
-                meta_data.append(col)
-                
-        columns_all=["activity"] + meta_data
-        df_new=self.df[columns_all].copy()
-        aggregated_df_first = df_new.groupby("activity").first().reset_index()
-        
-        return aggregated_df_first
-    @staticmethod
-    def df_to_kg(attributes_df, graph_df) -> nx:
+        timestamp = self.main_columns_dict["timestamp"]
+        activity= self.main_columns_dict["activity"]
+        case_id = self.main_columns_dict["case_id"]
+
+        def custom_agg(x):
+            if pd.api.types.is_numeric_dtype(x):
+                return x.mean()
+            elif pd.api.types.is_string_dtype(x):
+                most_common = Counter(x).most_common(3)
+                return [item for item, count in most_common]
+            else:
+                return list(x)
+    
+        grouped_df = self.df.drop(columns=[timestamp,case_id]).groupby(activity).agg(custom_agg).reset_index()
+        return grouped_df
+
+    def df_to_kg(self,attributes_df, graph_df) -> nx:
         """
     Create a knowledge graph from given dataframes and return the graph object.
     
@@ -83,13 +86,14 @@ class KgMaker:
     Returns:
     - G: NetworkX graph object
     """
+        activity= self.main_columns_dict["activity"]
         
         G = nx.DiGraph()
 
         for index, row in attributes_df.iterrows():
-            activity = row['Activity']
-            attributes = row.drop('Activity').to_dict()
-            G.add_node(activity, **attributes)
+            activity_temp = row[activity]
+            attributes = row.drop(activity).to_dict()
+            G.add_node(activity_temp, **attributes)
             
         for index, row in graph_df.iterrows():
             source = row['Source']