Optimize inference code

tutorials/model_ensembling.ipynb

@@ -883,39 +883,6 @@
     "### 5.1 Preparing the Data"
    ]
   },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "id": "553f4f32",
-   "metadata": {},
-   "outputs": [],
-   "source": [
-    "# rank_df = pd.DataFrame({\n",
-    "#     \"user_idx\": all_df[\"user_idx\"],\n",
-    "#     \"item_idx\": all_df[\"item_idx\"],\n",
-    "# })\n",
-    "\n",
-    "# total_items = train_set.num_items # 1651 items\n",
-    "\n",
-    "# models_to_calculate = [bpr_model, wmf_model, wmf_model_123, wmf_model_456, wmf_model_789, wmf_model_888, wmf_model_999, wmf_model_k20, wmf_model_k30, wmf_model_k40, wmf_model_k50]\n",
-    "\n",
-    "# # Calculate points for each model using the Borda count process.\n",
-    "# # Take note that points should be calculated on a per user basis.\n",
-    "# for model in tqdm(models_to_calculate):\n",
-    "#     name = model.name\n",
-    "#     rank_df[name + \"_score\"] = rank_df.apply(lambda row: model.score(int(row['user_idx']), int(row['item_idx'])), axis=1)\n",
-    "#     point_list = [] # This list will be added into the dataframe later\n",
-    "\n",
-    "#     # Calculate points for each user\n",
-    "#     for user_idx in range(train_set.num_users):\n",
-    "#         sub_rank_df = rank_df[rank_df[\"user_idx\"] == user_idx].copy() # get all items for a user\n",
-    "#         sub_rank_df.loc[:, name + \"_rank\"] = sub_rank_df[name + \"_score\"].rank(ascending=False).astype(int)  # Rank items\n",
-    "#         sub_rank_df.loc[:, name + \"_points\"] = total_items - sub_rank_df[name + \"_rank\"]  # Calculate points\n",
-    "#         point_list.extend(sub_rank_df[name + \"_points\"].values.tolist())  # Add points to the list\n",
-    "    \n",
-    "#     rank_df[name + \"_points\"] = point_list\n"
-   ]
-  },
   {
    "cell_type": "code",
    "execution_count": null,