2025/coursework/cw2/code_question/shapley_value_evaluation.py

@@ -18,10 +18,6 @@ def get_missing_steps(row, steps):
     ###Question 4: INSERT CODE HERE: Get missing steps as a tuple for each row.###
     ##############################################################################
     raise NotImplementedError("Get missing steps as a tuple for each row.")   
-    for step in steps:
-        col_name = f'step{step}_missing'
-        if row[col_name] == 1:
-            missing_steps.append(step)
     return tuple(sorted(missing_steps))
 
 def generate_all_subsets(steps):
@@ -53,13 +49,6 @@ def compute_v_S(df, all_subsets_missing):
     ###Question 5: INSERT CODE HERE: Compute v(S) for all subsets of missing steps.###
     ##################################################################################
     raise NotImplementedError("Implement the code to compute v(S) for all subsets of missing steps.")    
-    v_S = {}
-    grouped = df.groupby('missing_steps')['is_correct'].mean().to_dict()
-    for S in all_subsets_missing:
-        if S in grouped:
-            v_S[S] = grouped[S]
-        else:
-            v_S[S] = np.nan  # Code Explanation: Please do not remove. Assign NaN if the subset is not in the data
     return v_S
 
 def compute_marginal_contributions(steps, v_S):
@@ -87,15 +76,6 @@ def compute_marginal_contributions(steps, v_S):
             ###Question 6: INSERT CODE HERE: Retrieve S_i, S_i_union_i, S_i_sorted, S_i_union_i_sorted###
             #############################################################################################
             raise NotImplementedError("Implement the retrieval of S_i, S_i_union_i, and their sorted tuples.")
-            S_i = set(pi[:idx_i])  # Steps before i in the permutation
-            S_i_union_i = S_i.union({i})
-            # Code Explanation below: Please do not remove. Additional tips about the function: Compute missing steps for S_i and S_i_union_i
-            missing_S_i = total_steps_set - S_i
-            missing_S_i_union_i = total_steps_set - S_i_union_i
-            # Code Explanation below: Please do not remove. Convert to sorted tuples to match keys in v_S
-            missing_S_i_sorted = tuple(sorted(missing_S_i))
-            missing_S_i_union_i_sorted = tuple(sorted(missing_S_i_union_i))
-
             v_S_i = v_S.get(missing_S_i_sorted, np.nan)
             v_S_i_union_i = v_S.get(missing_S_i_union_i_sorted, np.nan)
             if np.isnan(v_S_i) or np.isnan(v_S_i_union_i):