binary_bpr

code/binary_bpr/main.py

@@ -26,8 +26,8 @@ from datetime import datetime
 
 dtype = torch.float32
 
-print(f"Is CUDA supported by this system? {torch.cuda.is_available()}")
-print(f"CUDA version: {torch.version.cuda}")
+#print(f"Is CUDA supported by this system? {torch.cuda.is_available()}")
+#print(f"CUDA version: {torch.version.cuda}")
 if torch.cuda.is_available():  
   dev = "cuda:0" 
 else:  
@@ -49,21 +49,21 @@ def read_file(dataTrain, dataTest):
     kc = flattern_arrays(kc.values, kcT.values)
     num_kc = len(kc)
     dico_kc = { k:v for (k,v) in zip(kc, range(len(kc)))}
-    print("NB KC", num_kc)
+    #print("NB KC", num_kc)
     # dico users
     users = df['user_id']
     usersT = dfTest['user_id']
     users = flattern_arrays(users.values, usersT.values)
     num_users = len(users)
     dico_users = { k:v for (k,v) in zip(users, range(num_users))}
-    print("NB Users", num_users)
+    #print("NB Users", num_users)
     # dico items and their associated kc
     itemsDT = df['item_id']
     itemsT = dfTest['item_id']
     items = flattern_arrays(itemsDT.values, itemsT.values)
     num_items = len(items)
     dico_items = { k:v for (k,v) in zip(items, range(num_items))} 
-    print("NB Items", num_items, len(dico_items))    
+    #print("NB Items", num_items, len(dico_items))    
     return dico_kc, dico_users, dico_items
 def save_embeddings(xpName: str, modelName: str, embeddings,userEmbDir : str,itemEmbDir : str, grid_search_id):
     """
@@ -76,7 +76,6 @@ def save_embeddings(xpName: str, modelName: str, embeddings,userEmbDir : str,ite
     """
 
     u_emb, i_emb = embeddings
-
     results_name_file = (xpName + modelName + "_" + str(grid_search_id))
 
     # save embeddings
@@ -90,8 +89,6 @@ def parse_dataframe(data, dico_kc, dico_users, dico_item, is_train = True):
     df = pd.read_csv(data, names=['user_id', 'item_id','correct','knowledge'])
     # Compute table of positive and negative items by KC and Users
     # and the dictionary that associate the KC to a question/answer
-    #num_kc = np.max(np.array(list(dico_kc.keys()))) + 1
-    #print("Parse DF", num_kc)
     num_kc = len(dico_kc)
     num_users = len(dico_users)
     # Find positive items for each kc/user
@@ -112,7 +109,6 @@ def parse_dataframe(data, dico_kc, dico_users, dico_item, is_train = True):
                 col = row['item_id']
                 if col not in dico_items:
                     dico_items[col] = len(dico_items)
-                # Warning, all user's answers are positives!
                 q,r = parse_it(col)
                 col_neg = q+'_'+str(1-int(r))
                 if col_neg not in dico_items:
@@ -155,7 +151,6 @@ def generate_quad(dico_items, triplets, t_trainy, item_users, alpha):
                     uu = item_users[t[2]][u]
                     t_quadriplets.append([t[0], t[1], t[2], uu])
                     t_y.append(t_trainy[k][i])
-                    #break
             else:
                 t_quadriplets.append([t[0], t[1], t[2], t[0]])
                 t_y.append(t_trainy[k][i])
@@ -278,7 +273,7 @@ if __name__ == '__main__':
     parser.add_argument("-bSize", "--batchSize", help="batch size")
     parser.add_argument("-lr", "--learningRate", help="learning rate")
     parser.add_argument("-mode", "--mode", help="CV mode = 1, GS mode = 0")
-    #parser.add_argument("-a", "--alpha", help="float")
+    
     args = parser.parse_args()
     dataTrain = args.dataTrain
     dataTest = args.dataTest
@@ -307,7 +302,6 @@ if __name__ == '__main__':
             FileNameTest_temp = testFileName[:-1] + str(i_fold)
             dataTrain = FileNameTrain_temp+".csv"
             dataTest = FileNameTest_temp+".csv"
-            # alpha = int(args.alpha)
             print("dataTrain:", dataTrain)
             print("dataTest:", dataTest)
             print("dataPath:", dataPath)
@@ -318,7 +312,6 @@ if __name__ == '__main__':
             dico_kc, dico_users, dico_items = read_file(dataTrain, dataTest)
             embedding_size = len(dico_kc)
             dico_items, t_train, ty_train, item_users = parse_dataframe(dataTrain, dico_kc, dico_users, dico_items, True)
-            # print("alpha", alpha)
             train, y_train = generate_quad(dico_items, t_train, ty_train, item_users, alpha)
             dico_items, test, y_test = parse_dataframe(dataTest, dico_kc, dico_users, dico_items, False)
             num_users = len(dico_users)
@@ -341,7 +334,7 @@ if __name__ == '__main__':
 
             write_file_doa(FileNameTrain_temp, emb[0], train, dico_kc, dico_users, dico_items)
             doa = compute_doa(FileNameTrain_temp)
-            # '''
+           
             # Test
             correctness, acc, users, auc, rmse = bpr_model.evaluate_model(test, len(dico_kc), y_test)
             acc_list.append(acc)
@@ -350,18 +343,13 @@ if __name__ == '__main__':
             doa_train.append(doa)
             print("Doa on Train dataset:", doa)
             print("AUC and RMSE on test dataset:", auc, rmse)
-            # '''
+           
             new_embedding_value = bpr_model.user_embeddings.weight.clone().detach().cpu().numpy()
             write_file_doa_test(FileNameTest_temp, new_embedding_value, test, y_test, dico_kc, dico_users, dico_items)
             doa = compute_doa(FileNameTest_temp)
             doa_test.append(doa)
             print("Accuracy and Doa on test dataset:", acc, doa)
 
-
-            # '''
-            ## test oppose
-            # acc, precision = bpr_model.evaluate_model(test1, len(dico_kc), y_test1)
-            # print(f'Accuracy: {acc}')
         print(acc_list)
         print(auc_list)
         print(rmse_list)
@@ -374,7 +362,6 @@ if __name__ == '__main__':
         print("doa_test :", np.mean(doa_test), "+-", np.std(doa_test))
         print("reo :",1- np.mean(doa_test)/np.mean(doa_train))
     else :
-        #alpha = int(args.alpha)
         print("dataTrain:",dataTrain)
         print("epochs:",epochs)
         print("batch_size:",batch_size)
@@ -382,7 +369,6 @@ if __name__ == '__main__':
         dico_kc, dico_users, dico_items = read_file(dataTrain, dataTest)
         embedding_size = len(dico_kc)
         dico_items, t_train, ty_train, item_users = parse_dataframe(dataTrain, dico_kc, dico_users, dico_items, True)
-        #print("alpha", alpha)
         train, y_train = generate_quad(dico_items, t_train, ty_train, item_users, alpha)
         dico_items, test, y_test = parse_dataframe(dataTest, dico_kc, dico_users, dico_items, False)
         num_users = len(dico_users)
@@ -403,16 +389,12 @@ if __name__ == '__main__':
         write_file_doa(trainFileName, emb[0], train, dico_kc, dico_users, dico_items)
         doa = compute_doa(trainFileName)
         print("Doa on train dataset:", doa)
-        #'''
         # Test
         correctness, acc, users, auc, rmse = bpr_model.evaluate_model(test, len(dico_kc), y_test)
         print(f'Accuracy: {acc}')
-        #'''
+        
         new_embedding_value = bpr_model.user_embeddings.weight.clone().detach().cpu().numpy()
         write_file_doa_test(testFileName, new_embedding_value, test, y_test, dico_kc, dico_users, dico_items)
         doa = compute_doa(testFileName)
         print("Accuracy and Doa on test dataset:", acc, doa)
-        #'''
-        ## test oppose
-        #acc, precision = bpr_model.evaluate_model(test1, len(dico_kc), y_test1)
-        #print(f'Accuracy: {acc}')
+        

code/binary_bpr/script.py

 import os
-dPath = "../../data/cdbpr_format/"
+dPath = "../../data/"
 embDirPath = "../../results/table_2/"
 datasets = ['assist0910_tkde', 'assist17_tkde', 'algebra','math_1', 'math_2']
 epochs = [75, 95, 5, 90, 90]